With the controlling party in Congress hanging in the balance, 51 percent of likely voters say they are extremely or very enthusiastic about voting for Congress this year; another 29 percent are somewhat enthusiastic while 19 percent are either not too or not at all enthusiastic. Today, Democrats and Republicans have about equal levels of enthusiasm, while independents are much less likely to be extremely or very enthusiastic. As Californians prepare to vote in the upcoming midterm election, fewer than half of adults and likely voters are satisfied with the way democracy is working in the United States—and few are very satisfied.

Satisfaction was higher in our February survey when 53 percent of adults and 48 percent of likely voters were satisfied with democracy in America. Today, half of Democrats and about four in ten independents are satisfied, compared to about one in five Republicans.

Notably, four in ten Republicans are not at all satisfied. In addition to the lack of satisfaction with the way democracy is working, Californians are divided about whether Americans of different political positions can still come together and work out their differences.

Forty-nine percent are optimistic, while 46 percent are pessimistic. Today, in a rare moment of bipartisan agreement, about four in ten Democrats, Republicans, and independents are optimistic that Americans of different political views will be able to come together. Notably, in , half or more across parties, regions, and demographic groups were optimistic. Today, about eight in ten Democrats—compared to about half of independents and about one in ten Republicans—approve of Governor Newsom.

Across demographic groups, about half or more approve of how Governor Newsom is handling his job. Approval of Congress among adults has been below 40 percent for all of after seeing a brief run above 40 percent for all of Democrats are far more likely than Republicans to approve of Congress.

Fewer than half across regions and demographic groups approve of Congress. Approval in March was at 44 percent for adults and 39 percent for likely voters.

Across demographic groups, about half or more approve among women, younger adults, African Americans, Asian Americans, and Latinos. Views are similar across education and income groups, with just fewer than half approving. Approval in March was at 41 percent for adults and 36 percent for likely voters.

Across regions, approval reaches a majority only in the San Francisco Bay Area. Across demographic groups, approval reaches a majority only among African Americans. This map highlights the five geographic regions for which we present results; these regions account for approximately 90 percent of the state population.

Residents of other geographic areas in gray are included in the results reported for all adults, registered voters, and likely voters, but sample sizes for these less-populous areas are not large enough to report separately.

The PPIC Statewide Survey is directed by Mark Baldassare, president and CEO and survey director at the Public Policy Institute of California. Coauthors of this report include survey analyst Deja Thomas, who was the project manager for this survey; associate survey director and research fellow Dean Bonner; and survey analyst Rachel Lawler.

The Californians and Their Government survey is supported with funding from the Arjay and Frances F. Findings in this report are based on a survey of 1, California adult residents, including 1, interviewed on cell phones and interviewed on landline telephones. The sample included respondents reached by calling back respondents who had previously completed an interview in PPIC Statewide Surveys in the last six months.

Interviews took an average of 19 minutes to complete. Interviewing took place on weekend days and weekday nights from October 14—23, Cell phone interviews were conducted using a computer-generated random sample of cell phone numbers.

Additionally, we utilized a registration-based sample RBS of cell phone numbers for adults who are registered to vote in California. All cell phone numbers with California area codes were eligible for selection. After a cell phone user was reached, the interviewer verified that this person was age 18 or older, a resident of California, and in a safe place to continue the survey e.

Cell phone respondents were offered a small reimbursement to help defray the cost of the call. Cell phone interviews were conducted with adults who have cell phone service only and with those who have both cell phone and landline service in the household. Landline interviews were conducted using a computer-generated random sample of telephone numbers that ensured that both listed and unlisted numbers were called.

Additionally, we utilized a registration-based sample RBS of landline phone numbers for adults who are registered to vote in California. All landline telephone exchanges in California were eligible for selection. For both cell phones and landlines, telephone numbers were called as many as eight times.

When no contact with an individual was made, calls to a number were limited to six. Also, to increase our ability to interview Asian American adults, we made up to three additional calls to phone numbers estimated by Survey Sampling International as likely to be associated with Asian American individuals.

Accent on Languages, Inc. The survey sample was closely comparable to the ACS figures. To estimate landline and cell phone service in California, Abt Associates used state-level estimates released by the National Center for Health Statistics—which used data from the National Health Interview Survey NHIS and the ACS. The estimates for California were then compared against landline and cell phone service reported in this survey.

We also used voter registration data from the California Secretary of State to compare the party registration of registered voters in our sample to party registration statewide. The sampling error, taking design effects from weighting into consideration, is ±3. This means that 95 times out of , the results will be within 3.

The sampling error for unweighted subgroups is larger: for the 1, registered voters, the sampling error is ±4. For the sampling errors of additional subgroups, please see the table at the end of this section.

Sampling error is only one type of error to which surveys are subject. Results may also be affected by factors such as question wording, question order, and survey timing. We present results for five geographic regions, accounting for approximately 90 percent of the state population. Residents of other geographic areas are included in the results reported for all adults, registered voters, and likely voters, but sample sizes for these less-populous areas are not large enough to report separately.

We also present results for congressional districts currently held by Democrats or Republicans, based on residential zip code and party of the local US House member. We compare the opinions of those who report they are registered Democrats, registered Republicans, and no party preference or decline-to-state or independent voters; the results for those who say they are registered to vote in other parties are not large enough for separate analysis.

We also analyze the responses of likely voters—so designated per their responses to survey questions about voter registration, previous election participation, intentions to vote this year, attention to election news, and current interest in politics.

The percentages presented in the report tables and in the questionnaire may not add to due to rounding. Additional details about our methodology can be found at www. pdf and are available upon request through surveys ppic.

October 14—23, 1, California adult residents; 1, California likely voters English, Spanish. Margin of error ±3. Percentages may not add up to due to rounding. Overall, do you approve or disapprove of the way that Gavin Newsom is handling his job as governor of California? Overall, do you approve or disapprove of the way that the California Legislature is handling its job?

Do you think things in California are generally going in the right direction or the wrong direction? Thinking about your own personal finances—would you say that you and your family are financially better off, worse off, or just about the same as a year ago?

Next, some people are registered to vote and others are not. Are you absolutely certain that you are registered to vote in California? Are you registered as a Democrat, a Republican, another party, or are you registered as a decline-to-state or independent voter? Would you call yourself a strong Republican or not a very strong Republican? Do you think of yourself as closer to the Republican Party or Democratic Party? Which one of the seven state propositions on the November 8 ballot are you most interested in?

Initiative Constitutional Amendment and Statute. It allows in-person sports betting at racetracks and tribal casinos, and requires that racetracks and casinos that offer sports betting to make certain payments to the state—such as to support state regulatory costs. The fiscal impact is increased state revenues, possibly reaching tens of millions of dollars annually.

Some of these revenues would support increased state regulatory and enforcement costs that could reach the low tens of millions of dollars annually. If the election were held today, would you vote yes or no on Proposition 26? Initiative Constitutional Amendment. It allows Indian tribes and affiliated businesses to operate online and mobile sports wagering outside tribal lands.

It directs revenues to regulatory costs, homelessness programs, and nonparticipating tribes. Some revenues would support state regulatory costs, possibly reaching the mid-tens of millions of dollars annually. If the election were held today, would you vote yes or no on Proposition 27? Initiative Statute. It allocates tax revenues to zero-emission vehicle purchase incentives, vehicle charging stations, and wildfire prevention.

If the election were held today, would you vote yes or no on Proposition 30? Do you agree or disagree with these statements? Overall, do you approve or disapprove of the way that Joe Biden is handling his job as president? Overall, do you approve or disapprove of the way Alex Padilla is handling his job as US Senator? Overall, do you approve or disapprove of the way Dianne Feinstein is handling her job as US Senator? Overall, do you approve or disapprove of the way the US Congress is handling its job?

Do you think things in the United States are generally going in the right direction or the wrong direction? How satisfied are you with the way democracy is working in the United States? Are you very satisfied, somewhat satisfied, not too satisfied, or not at all satisfied? These days, do you feel [rotate] [1] optimistic [or] [2] pessimistic that Americans of different political views can still come together and work out their differences? What is your opinion with regard to race relations in the United States today?

Would you say things are [rotate 1 and 2] [1] better , [2] worse , or about the same than they were a year ago? When it comes to racial discrimination, which do you think is the bigger problem for the country today—[rotate] [1] People seeing racial discrimination where it really does NOT exist [or] [2] People NOT seeing racial discrimination where it really DOES exist?

Next, Next, would you consider yourself to be politically: [read list, rotate order top to bottom]. Generally speaking, how much interest would you say you have in politics—a great deal, a fair amount, only a little, or none?

Mark Baldassare is president and CEO of the Public Policy Institute of California, where he holds the Arjay and Frances Fearing Miller Chair in Public Policy.

He is a leading expert on public opinion and survey methodology, and has directed the PPIC Statewide Survey since He is an authority on elections, voter behavior, and political and fiscal reform, and the author of ten books and numerous publications. Before joining PPIC, he was a professor of urban and regional planning in the School of Social Ecology at the University of California, Irvine, where he held the Johnson Chair in Civic Governance.

He has conducted surveys for the Los Angeles Times , the San Francisco Chronicle , and the California Business Roundtable. He holds a PhD in sociology from the University of California, Berkeley. Dean Bonner is associate survey director and research fellow at PPIC, where he coauthors the PPIC Statewide Survey—a large-scale public opinion project designed to develop an in-depth profile of the social, economic, and political attitudes at work in California elections and policymaking.

He has expertise in public opinion and survey research, political attitudes and participation, and voting behavior. Before joining PPIC, he taught political science at Tulane University and was a research associate at the University of New Orleans Survey Research Center. He holds a PhD and MA in political science from the University of New Orleans. Rachel Lawler is a survey analyst at the Public Policy Institute of California, where she works with the statewide survey team.

In that role, she led and contributed to a variety of quantitative and qualitative studies for both government and corporate clients. She holds an MA in American politics and foreign policy from the University College Dublin and a BA in political science from Chapman University. Deja Thomas is a survey analyst at the Public Policy Institute of California, where she works with the statewide survey team. Inspection points and meta-abduction in logic programs. A survey of first-order probabilistic models.

A distributed architecture for norm-aware agent societies. Strong equivalence of RASP programs. Challenges in relational learning for real-time systems applications. From core foundational and theoretical material to final-year topics and applications, UTiCS books take a fresh, concise, and modern approach and are ideal for self-study or for a one- or two-semester course.

The texts are all au- thored by established experts in their fields, reviewed by an international advisory board, and contain numerous examples and problems. For further volumes: www. Wolfgang Ertel FB Elektrotechnik und Informatik Hochschule Ravensburg-Weingarten University of Applied Sciences Weingarten Germany ertel hs-weingarten.

de Series editor Ian Mackie Advisory board Samson Abramsky, University of Oxford, Oxford, UK Karin Breitman, Pontifical Catholic University of Rio de Janeiro, Rio de Janeiro, Brazil Chris Hankin, Imperial College London, London, UK Dexter Kozen, Cornell University, Ithaca, USA Andrew Pitts, University of Cambridge, Cambridge, UK Hanne Riis Nielson, Technical University of Denmark, Kongens Lyngby, Denmark Steven Skiena, Stony Brook University, Stony Brook, USA Iain Stewart, University of Durham, Durham, UK ISSN ISBN e-ISBN DOI Enquiries concerning reproduction outside those terms should be sent to the publishers.

The use of registered names, trademarks, etc. The publisher makes no representation, express or implied, with regard to the accuracy of the information contained in this book and cannot accept any legal responsibility or liability for any errors or omissions that may be made. However, the methods and formalisms used on the way to this goal are not firmly set, which has resulted in AI consisting of a multitude of subdisciplines today. The difficulty in an introductory AI course lies in conveying as many branches as possible without losing too much depth and precision.

However, since this book has 1, pages, and since it is too extensive and costly for most students, the requirements for writing this book were clear: it should be an accessible introduction to modern AI for self-study or as the foundation of a four-hour lecture, with at most pages. The result is in front of you. In the space of pages, a field as extensive as AI cannot be fully covered. To avoid turning the book into a table of contents, I have attempted to go into some depth and to introduce concrete algorithms and applications in each of the following branches: agents, logic, search, reasoning with uncertainty, machine learning, and neural networks.

The fields of image processing, fuzzy logic, and natural language processing are not covered in detail. The field of image processing, which is important for all of computer science, is a stand-alone discipline with very good textbooks, such as [GW08]. Natural language processing has a similar status. In recognizing and gen- erating text and spoken language, methods from logic, probabilistic reasoning, and neural networks are applied.

In this sense this field is part of AI. On the other hand, computer linguistics is its own extensive branch of computer science and has much in common with formal languages. In this book we will point to such appropriate systems in several places, but not give a systematic introduction. For a first introduc- tion in this field, we refer to Chaps. Fuzzy logic, or fuzzy set theory, has developed into a branch of control theory due to its primary application in automation technology and is covered in the corresponding books and lectures.

Therefore we will forego an introduction here. The dependencies between chapters of the book are coarsely sketched in the graph shown below. To keep it simple, Chap. As an example, the thicker arrow from 2 to 3 means that propositional logic is a prerequisite for understanding predicate logic. The thin arrow from 9 to 10 means that neural networks are helpful for un- derstanding reinforcement learning, but not absolutely necessary.

This book is applicable to students of computer science and other technical natural sciences and, for the most part, requires high school level knowledge of mathemat- ics. In several places, knowledge from linear algebra and multidimensional analysis is needed. For a deeper understanding of the contents, actively working on the ex- ercises is indispensable.

I ask the reader to please send suggestions, criticisms, and tips about errors directly to ertel hs-weingarten.

My special thanks go to the translator Nathan Black who in an excellent trans-Atlantic cooperation between Germany and California via SVN, Skype and Email produced this text.

I am grateful to Franz Kur- feß, who introduced me to Nathan; to Matthew Wight for proofreading the translated book and to Simon Rees from Springer Verlag for his patience.

I would like to thank my wife Evelyn for her support and patience during this time consuming project. Special thanks go to Wolfgang Bibel and Chris Loben- schuss, who carefully corrected the German manuscript. Their suggestions and dis- cussions lead to many improvements and additions. For reading the corrections and other valuable services, I would like to thank Richard Cubek, Celal Döven, Joachim Feßler, Nico Hochgeschwender, Paul Kirner, Wilfried Meister, Norbert Perk, Pe- ter Radtke, Markus Schneider, Manfred Schramm, Uli Stärk, Michel Tokic, Arne Usadel and all interested students.

My thanks also go out to Florian Mast for the priceless cartoons and very effective collaboration. I hope that during your studies this book will help you share my fascination with Artificial Intelligence. The term artificial intelligence stirs emotions. For one thing there is our fascination with intelligence, which seemingly imparts to us humans a special place among life forms. All these questions are meaningful when trying to understand artificial intelligence.

However, the central question for the engineer, especially for the computer scientist, is the question of the intelligent machine that behaves like a person, showing intelligent behavior. The attribute artificial might awaken much different associations. It brings up fears of intelligent cyborgs.

It recalls images from science fiction novels. It raises the question of whether our highest good, the soul, is something we should try to understand, model, or even reconstruct. With such different offhand interpretations, it becomes difficult to define the term artificial intelligence or AI simply and robustly. Nevertheless I would like to try, using examples and historical definitions, to characterize the field of AI.

In , John McCarthy, one of the pioneers of AI, was the first to define the term artificial intelligence, roughly as follows: The goal of AI is to develop machines that behave as though they were intelligent. To test this definition, the reader might imagine the following scenario. Fif- teen or so small robotic vehicles are moving on an enclosed four by four meter square surface. One can observe various behavior patterns. Some vehicles form small groups with relatively little movement. Others move peacefully through the space and gracefully avoid any collision.

Still others appear to follow a leader. Ag- gressive behaviors are also observable. Is what we are seeing intelligent behavior? The psychologist Valentin Braitenberg has shown that this seemingly complex behavior can be produced by very simple electrical circuits [Bra84]. So- called Braitenberg vehicles have two wheels, each of which is driven by an inde- pendent electric motor. The speed of each motor is influenced by a light sensor on W.

Ertel, Introduction to Artificial Intelligence, 1 Undergraduate Topics in Computer Science, DOI The more light that hits the sensor, the faster the motor runs. Vehicle 1 in the left part of the figure, according to its configuration, moves away from a point light source.

Vehicle 2 on the other hand moves toward the light source. Further small modifications can create other behav- ior patterns, such that with these very simple vehicles we can realize the impressive behavior described above. Clearly the above definition is insufficient because AI has the goal of solving difficult practical problems which are surely too demanding for the Braitenberg ve- hicle. In the Encyclopedia Britannica [Bri91] one finds a Definition that goes like: AI is the ability of digital computers or computer controlled robots to solve problems that are normally associated with the higher intellectual processing capabilities of humans.

But this definition also has weaknesses. It would admit for example that a com- puter with large memory that can save a long text and retrieve it on demand displays intelligent capabilities, for memorization of long texts can certainly be considered a higher intellectual processing capability of humans, as can for example the quick multiplication of two digit numbers.

According to this definition, then, every computer is an AI system. This dilemma is solved elegantly by the following defi- nition by Elaine Rich [Ric83]: Artificial Intelligence is the study of how to make computers do things at which, at the moment, people are better.

Rich, tersely and concisely, characterizes what AI researchers have been doing for the last 50 years. Even in the year , this definition will be up to date. Tasks such as the execution of many computations in a short amount of time are the strong points of digital computers. In this regard they outperform humans by many multiples. In many other areas, however, humans are far superior to machines.

For instance, a person entering an unfamiliar room will recognize the surroundings within fractions of a second and, if necessary, just as swiftly make decisions and plan actions. To date, this task is too demanding for autonomous1 robots. In fact, research on autonomous robots is an important, current theme in AI. Construction of chess computers, on the other hand, has lost relevance because they already play at or above the level of grandmasters. This also shows that the other cited definitions reflect important aspects of AI.

A particular strength of human intelligence is adaptivity. We are capable of ad- justing to various environmental conditions and change our behavior accordingly through learning. Many ideas and princi- ples in the field of neural networks see Chap. A very different approach results from taking a goal-oriented line of action, start- ing from a problem and trying to find the most optimal solution. How humans solve the problem is treated as unimportant here.

The method, in this approach, is sec- ondary. First and foremost is the optimal intelligent solution to the problem. Rather than employing a fixed method such as, for example, predicate logic AI has as its constant goal the creation of intelligent agents for as many different tasks as pos- sible.

Because the tasks may be very different, it is unsurprising that the methods currently employed in AI are often also quite different. Similar to medicine, which encompasses many different, often life-saving diagnostic and therapy procedures, AI also offers a broad palette of effective solutions for widely varying applications.

For mental inspiration, consider Fig. Just as in medicine, there is no universal method for all application areas of AI, rather a great number of possible solutions for the great number of various everyday problems, big and small. Cognitive science is devoted to research into human thinking at a somewhat higher level. Similarly to brain science, this field furnishes practical AI with many important ideas. On the other hand, algorithms and implementations lead to fur- ther important conclusions about how human reasoning functions.

Thus these three fields benefit from a fruitful interdisciplinary exchange. The subject of this book, however, is primarily problem-oriented AI as a subdiscipline of computer science.

There are many interesting philosophical questions surrounding intelligence and artificial intelligence. We humans have consciousness; that is, we can think about ourselves and even ponder that we are able to think about ourselves.

How does consciousness come to be? Many philosophers and neurologists now believe that the mind and consciousness are linked with matter, that is, with the brain. The question of whether machines could one day have a mind or consciousness could at some point in the future become relevant.

The mind-body problem in particular concerns whether or not the mind is bound to the body. We will not discuss these questions here.

The interested reader may consult [Spe98, Spe97] and is invited, in the course of AI technology studies, to form a personal opinion about these questions. The test person Alice sits in a locked room with two computer terminals. One terminal is connected to a machine, the other with a non-malicious person Bob. Alice can type questions into both terminals. She is given the task of deciding, after five minutes, which terminal belongs to the machine.

The reasons for this are similar to those mentioned above related to Braitenberg vehicles see Exercise 1. He was in fact able to demonstrate success in many cases. Supposedly his secretary often had long discussions with the program. Today in the internet there are many so-called chatterbots, some of whose initial responses are quite impres- sive. After a certain amount of time, however, their artificial nature becomes appar- ent.

Some of these programs are actually capable of learning, while others possess extraordinary knowledge of various subjects, for example geography or software development. There are already commercial applications for chatterbots in online customer support and there may be others in the field of e-learning. It is conceivable that the learner and the e-learning system could communicate through a chatterbot. The reader may wish to compare several chatterbots and evaluate their intelligence in Exercise 1.

Table 1. The completeness theorem states that first-order predicate logic is complete. This means that every true statement that can be formulated in predi- cate logic is provable using the rules of a formal calculus. On this basis, automatic theorem provers could later be constructed as implementations of formal calculi. With the incompleteness theorem, Gödel showed that in higher-order logics there exist true statements that are unprovable.

He showed that there is no program that can decide whether a given arbitrary program and its respective input will run in an infinite loop.

With 2 Higher-orderlogics are extensions of predicate logic, in which not only variables, but also func- tion symbols or predicates can appear as terms in a quantification. Indeed, Gödel only showed that any system that is based on predicate logic and can formulate Peano arithmetic is incomplete.

It follows, for example, that there will never be a universal program verification system. However, computers at that time lacked sufficient power to simulate simple brains. This was the case in the s. Newell and Simon introduced Logic Theorist, the first automatic theorem prover, and thus also showed that with computers, which actually only work with numbers, one can also process symbols.

At the same time McCarthy introduced, with the language LISP, a programming language specially created for the processing of symbolic structures. Both of these systems were introduced in at the historic Dartmouth Confer- ence, which is considered the birthday of AI. In the US, LISP developed into the most important tool for the implementation of symbol-processing AI systems.

Thereafter the logical inference rule known as resolution developed into a complete calculus for predicate logic. PROLOG offers the advantage of allowing direct programming using Horn clauses, a subset of predicate logic. Like LISP, PROLOG has data types for convenient processing of lists. Until well into the s, a breakthrough spirit dominated AI, especially among many logicians. The reason for this was the string of impressive achievements in symbol processing.

With the Fifth Generation Computer Systems project in Japan and the ESPRIT program in Europe, heavy investment went into the construction of intelligent computers. For small problems, automatic provers and other symbol-processing systems sometimes worked very well. The combinatorial explosion of the search space, how- ever, defined a very narrow window for these successes.

Because the economic success of AI systems fell short of expectations, funding for logic-based AI research in the United States fell dramatically during the s. Because of the fault-tolerance of such systems and their ability to recognize pat- terns, considerable successes became possible, especially in pattern recognition.

Facial recognition in photos and handwriting recognition are two example appli- cations. The system Nettalk was able to learn speech from example texts [SR86]. Under the name connectionism, a new subdiscipline of AI was born. Connectionism boomed and the subsidies flowed.

But soon even here feasibility limits became obvious. The neural networks could acquire impressive capabilities, but it was usually not possible to capture the learned concept in simple formulas or logical rules. Attempts to combine neural nets with logical rules or the knowledge of human experts met with great difficulties.

Additionally, no satisfactory solution to the structuring and modularization of the networks was found. Several alternatives were suggested. The most promising, probabilistic reasoning, works with conditional probabili- ties for propositional calculus formulas. Since then many diagnostic and expert sys- tems have been built for problems of everyday reasoning using Bayesian networks.

The weaknesses of logic, which can only work with two truth values, can be solved by fuzzy logic, which pragmatically introduces infinitely many values be- tween zero and one. Though even today its theoretical foundation is not totally firm, it is being successfully utilized, especially in control engineering.

A much different path led to the successful synthesis of logic and neural networks under the name hybrid systems. For example, neural networks were employed to learn heuristics for reduction of the huge combinatorial search space in proof dis- covery [SE90].

Methods of decision tree learning from data also work with probabilities. Sys- tems like CART, ID3 and C4. Today they are a favorite among machine learning techniques Sect.

Since about , data mining has developed as a subdiscipline of AI in the area of statistical data analysis for extraction of knowledge from large databases. Data mining brings no new techniques to AI, rather it introduces the requirement of us- ing large databases to gain explicit knowledge.

One application with great market potential is steering ad campaigns of big businesses based on analysis of many mil- lions of purchases by their customers.

Typically, machine learning techniques such as decision tree learning come into play here. One of its goals is the use of parallel computers to increase the efficiency of problem solvers. A very different conceptual approach results from the development of au- tonomous software agents and robots that are meant to cooperate like human teams. As with the aforementioned Braitenberg vehicles, there are many cases in which an individual agent is not capable of solving a problem, even with unlimited re- sources.

Only the cooperation of many agents leads to the intelligent behavior or to the solution of a problem. An ant colony or a termite colony is capable of erecting buildings of very high architectural complexity, despite the fact that no single ant comprehends how the whole thing fits together.

This is similar to the situation of provisioning bread for a large city like New York [RN10]. There is no central plan- ning agency for bread, rather there are hundreds of bakers that know their respective areas of the city and bake the appropriate amount of bread at those locations. Active skill acquisition by robots is an exciting area of current research. There are robots today, for example, that independently learn to walk or to perform various motorskills related to soccer Chap. Cooperative learning of multiple robots to solve problems together is still in its infancy.

Most of these tools are well-developed and are available as finished software libraries, often with con- venient user interfaces. The selection of the right tool and its sensible use in each individual case is left to the AI developer or knowledge engineer. Like any other artisanship, this requires a solid education, which this book is meant to promote.

More than nearly any other science, AI is interdisciplinary, for it draws upon interesting discoveries from such diverse fields as logic, operations research, statis- tics, control engineering, image processing, linguistics, philosophy, psychology, and neurobiology.

On top of that, there is the subject area of the particular application. To successfully develop an AI project is therefore not always so simple, but almost always extremely exciting. Agent denotes rather generally a system that processes information and produces an output from an input. These agents may be classified in many different ways. In classical computer science, software agents are primarily employed Fig.

In this case the agent consists of a program that calculates a result from user input. In robotics, on the other hand, hardware agents also called robots are employed, which additionally have sensors and actuators at their disposal Fig. The agent can perceive its environment with the sensors. With the actuators it carries out actions and changes its environment. With respect to the intelligence of the agent, there is a distinction between reflex agents, which only react to input, and agents with memory, which can also include the past in their decisions.

For example, a driving robot that through its sensors knows its exact position and the time has no way, as a reflex agent, of determining its velocity.

If, however, it saves the position, at short, discrete time steps, it can thus easily calculate its average velocity in the previous time interval. If a reflex agent is controlled by a deterministic program, it represents a function of the set of all inputs to the set of all outputs. An agent with memory, on the other hand, is in general not a function.

See Exercise 1. Reflex agents are sufficient in cases where the problem to be solved involves a Markov decision process. This is a process in which only the current state is needed to determine the optimal next action see Chap. A mobile robot which should move from room to room in a building takes actions different from those of a robot that should move to room In other words, the actions depend on the goal.

Such agents are called goal-based. In higher-order logics, on the other hand, there are true statements that are unprovable [Göd31b]. In [Göd31b] Gödel showed that predicate logic extended with the axioms of arithmetic is incomplete. With vacuum tubes he simulates 40 neurons. Here the name Artificial Intelligence was first introduced. Newell and Simon of Carnegie Mellon University CMU present the Logic Theorist, the first symbol-processing computer program [NSS83].

He writes programs that are capable of modifying themselves. It goes unnoticed in the mainstream AI community of the time Sect. The system Nettalk learns to read texts aloud [SR86] Chap. Multi-agent systems become popular. First international RoboCup competition in Japan. Its goal as a goal- based agent is to put all emails in the right category. In the course of this not-so- simple task, the agent can occasionally make mistakes.

Because its goal is to classify all emails correctly, it will attempt to make as few errors as possible. However, that is not always what the user has in mind. Let us compare the following two agents. Out of 1, emails, Agent 1 makes only 12 errors.

Agent 2 on the other hand makes 38 errors with the same 1, emails. Is it therefore worse than Agent 1? Because there are in this case two types of errors of differing severity, each error should be weighted with the appropriate cost factor see Sect. The sum of all weighted errors gives the total cost caused by erroneous decisions. The goal of a cost-based agent is to minimize the cost of erroneous decisions in the long term, that is, on average.

In Sect. Analogously, the goal of a utility-based agent is to maximize the utility derived from correct decisions in the long term, that is, on average. The sum of all decisions weighted by their respective utility factors gives the total utility.

Of particular interest in AI are Learning agents, which are capable of changing themselves given training examples or through positive or negative feedback, such that the average utility of their actions grows over time see Chap. As mentioned in Sect. The design of an agent is oriented, along with its objective, strongly toward its environment, or alternately its picture of the environment, which strongly depends on it sensors.

The environment is observable if the agent always knows the com- plete state of the world. Otherwise the environment is only partially observable. If an action always leads to the same result, then the environment is deterministic. Otherwise it is nondeterministic. In a discrete environment only finitely many states and actions occur, whereas a continuous environment boasts infinitely many states or actions. For simple agents this way of looking at the problem is sufficient.

For complex applica- tions in which the agent must be able to rely on a large amount of information and is meant to do a difficult task, programming the agent can be very costly and unclear how to proceed. Here AI provides a clear path to follow that will greatly simplify the work. First we separate knowledge from the system or program, which uses the knowl- edge to, for example, reach conclusions, answer queries, or come up with a plan.

This system is called the inference mechanism. The knowledge is stored in a knowl- edge base KB. Acquisition of knowledge in the knowledge base is denoted Knowl- edge Engineering and is based on various knowledge sources such as human experts, the knowledge engineer, and databases. Active learning systems can also acquire knowledge though active exploration of the world see Chap.

In Fig. Moving toward a separation of knowledge and inference has several crucial ad- vantages. The separation of knowledge and inference can allow inference systems to be implemented in a largely application-independent way. Through the decoupling of the knowledge base from inference, knowledge can be stored declaratively. In the knowledge base there is only a description of the knowledge, which is independent from the inference system in use.

Without this clear separation, knowledge and processing of inference steps would be interwoven, and any changes to the knowledge would be very costly. Formal language as a convenient interface between man and machine lends itself to the representation of knowledge in the knowledge base. In the following chap- ters we will get to know a whole series of such languages.

First, in Chaps. But other for- malisms such as probabilistic logic, fuzzy logic or decision trees are also presented. We start with propositional calculus and the related inference systems. Building on that, we will present predicate logic, a powerful language that is accessible by ma- chines and very important in AI.

Developed at IBM together with a number of universities, Watson is a question answering program, that can be fed with clues given in natural language. In the U. The high performance and short reaction times of Watson were due to an im- plementation on 90 IBM Power servers, each of which contains 32 processors, resulting in parallel processors. Start for exam- ple with www. com or www. Write down a starting question and measure the time it takes, for each of the various programs, until you know for certain that it is not a human.

com you will find a server on which you can build a chatterbot with the markup language AIML quite easily. Depending on your interest level, develop a simple or complex chatterbot, or change an existing one.

Exercise 1. are NP-complete or even undecidable. What does this mean for AI? b How can one change the agent with memory, or model it, such that it becomes equivalent to a function but does not lose its memory? b How must the agent be changed so that it can also calculate its acceleration? Provide a formula here as well. Assume here that having to manually delete a spam email costs one cent and retrieving a deleted email, or the loss of an email, costs one dollar. b Determine for both agents the profit created by correct classifications and com- pare the results.

Assume that for every desired email recognized, a profit of one dollar accrues and for every correctly deleted spam email, a profit of one cent. Propositional Logic 2 In propositional logic, as the name suggests, propositions are connected by logical operators. These two propositions can be connected to form the new proposition if it is raining the street is wet. This notation has the advantage that the elemental propositions appear again in un- altered form.

So that we can work with propositional logic precisely, we will begin with a definition of the set of all propositional logic formulas. The sets Op, Σ and {t, f } are pairwise disjoint.

Σ is called the signature and its elements are the proposition variables. This elegant recursive definition of the set of all formulas allows us to generate infinitely many formulas. Ertel, Introduction to Artificial Intelligence, 15 Undergraduate Topics in Computer Science, DOI We are still missing the semantics. The answer is: it depends on whether the variables A and B are true.

We must obviously assign truth values that reflect the state of the world to propo- sition variables. Therefore we define Definition 2. Because every proposition variable can take on two truth values, every proposi- tional logic formula with n different variables has 2n different interpretations.

We define the truth values for the basic operations by showing all possible interpreta- tions in a truth table see Table 2. The empty formula is true for all interpretations. In order to determine the truth value for complex formulas, we must also define the order of operations for logical operators. To clearly differentiate between the equivalence of formulas and syntactic equiv- alence, we define Definition 2. Semantic equivalence serves above all to be able to use the meta-language, that is, natural language, to talk about the object language, namely logic.

According to how many interpretations in which a formula is true, we can divide formulas into the following classes: Definition 2.

True for- mulas are also called tautologies. Every interpretation that satisfies a formula is called a model of the formula. Clearly the negation of every generally valid formula is unsatisfiable. The nega- tion of a satisfiable, but not generally valid formula F is satisfiable. We are now able to create truth tables for complex formulas to ascertain their truth values. We put this into action immediately using equivalences of formulas which are important in practice.

In propositional logic this means showing that a knowledge base KB—that is, a possibly extensive propositional logic formula— a formula Q1 follows. In other words, in every interpretation in which KB is true, Q is also true. More succinctly, whenever KB is true, Q is also true. Because, for the concept of entail- ment, interpretations of variables are brought in, we are dealing with a semantic concept.

Every formula that is not valid chooses so to speak a subset of the set of all inter- pretations as its model. The empty formula is therefore true in all interpretations.

Intuitively this means that tautologies are always true, without restriction of the in- terpretations by a formula. Now we show an important connection between the semantic concept of entailment and syntactic implication. Theorem 2. This means that for every interpretation that makes A true, B is also true. The critical second row of the truth table does not even apply in that case.

Thus one direction of the statement has been shown. Thus the critical second row of the truth table is also locked out. Every model of A is then also a model of B. Thus we have our first proof system for propositional logic, which is easily automated. The disadvantage of this method is the very long computation time in the worst case. Therefore this process is unusable for large variable counts, at least in the worst case.

We formulate this simple, but important consequence of the deduction theorem as a theorem. To show that the query Q follows from the knowledge base KB, we can also add the negated query ¬Q to the knowledge base and derive a contradiction. Therefore, Q has been proved. This procedure, which is frequently used in mathematics, is also used in various automatic proof calculi such as the resolution calculus and in the processing of PROLOG programs.

Such syntactic proof systems are called calculi. To ensure that a calculus does not gener- ate errors, we define two fundamental properties of calculi. Definition 2. A calculus is called complete if all semantic consequences can be derived.

The soundness of a calculus ensures that all derived formulas are in fact seman- tic consequences of the knowledge base. The completeness of a calculus, on the other hand, ensures that the calculus does not overlook anything. A complete calculus always finds a proof if the formula to be proved follows from the knowledge base. Mod X represents the set of models of a formula X and complete, then syntactic derivation and semantic entailment are two equivalent relations see Fig.

To keep automatic proof systems as simple as possible, these are usually made to operate on formulas in conjunctive normal form.

Finally, a literal is a variable positive literal or a negated variable negative literal. The conjunctive normal form does not place a restriction on the set of formulas because: Theorem 2.

B This notation means that we can derive the formula s below the line from the comma-separated formulas above the line. Modus ponens as a rule by itself, while sound, is not complete. If we add additional rules we can create a complete calcu- lus, which, however, we do not wish to consider here.

The derived clause is called resolvent. The resolution rule is equally usable if C is missing or if A and C are missing. With the literals A1 ,. The resolution rule deletes a pair of complementary literals from the two clauses and combines the rest of the literals into a new clause. To prove that from a knowledge base KB, a query Q follows, we carry out a proof by contradiction. Following Theorem 2. In formulas in conjunctive normal form, a contradiction appears in the form of two clauses A and ¬A , which lead to the empty clause as their resolvent.

The following theorem ensures us that this process really works as desired. For the calculus to be complete, we need a small addition, as shown by the fol- lowing example. With the resolution rule alone, this is impossible. With factorization, which allows deletion of copies of lit- erals from clauses, this problem is eliminated. Otherwise anything can be derived from KB see Exercise 2. This is true not only of resolution, but also for many other calculi. Of the calculi for automated deduction, resolution plays an exceptional role.

Thus we wish to work a bit more closely with it. In contrast to other calculi, resolution has only two inference rules, and it works with formulas in conjunctive normal form. This makes its implementation simpler. A further advantage compared to many cal- culi lies in its reduction in the number of possibilities for the application of inference rules in every step of the proof, whereby the search space is reduced and computa- tion time decreased. As an example, we start with a simple logic puzzle that allows the important steps of a resolution proof to be shown.

Example 2. Recently, moved by noble feelings, I picked up three hitchhikers, a father, mother, and daughter, who I quickly realized were English and only spoke English. At each of the sentences that follow I wa- vered between two possible interpretations. To solve this kind of problem we proceed in three steps: formalization, trans- formation into normal form, and proof. In many cases formalization is by far the most difficult step because it is easy to make mistakes or forget small details.

Thus practical exercise is very important. See Exercises 2. Factoring out ¬S in the middle sub-formula brings the formula into CNF in one step. Now we begin the resolution proof, at first still without a query Q.

Every further resolution step would lead to the derivation of clauses that are already available. Be- cause it does not allow the derivation of the empty clause, it has therefore been shown that the knowledge base is non-contradictory. So far we have derived N and P. To show that ¬S holds, we add the clause S 8 to the set of clauses as a negated query. With the resolution step Res 2, 8 : 9 the proof is complete.

The bar is set to 1. If the second girl said the same to the third, who in turn said the same to the first, would it be possible for all three to win their bets? We show through proof by resolution that not all three can win their bets.

This implication contains the premise, a conjunction of variables and the conclusion, a disjunction of variables. The receiver of this message knows for certain that the sender is not going swimming.

The receiver now knows definitively. Thus we call clauses with at most one positive literal definite clauses. These clauses have the advantage that they only allow one conclusion and are thus distinctly simpler to interpret. Many relations can be described by clauses of this type. We therefore define Definition 2.

are named Horn clauses after their inventor. A clause with a single positive literal is a fact. In clauses with negative and one positive literal, the positive literal is called the head. To better understand the representation of Horn clauses, the reader may derive them from the definitions of the equivalences we have currently been using Exer- cise 2.

Horn clauses are easier to handle not only in daily life, but also in formal reason- ing, as we can see in the following example. With modus ponens we obtain a complete calculus for formulas that consist of propositional logic Horn clauses. In the case of large knowledge bases, however, modus ponens can derive many unnecessary formulas if one begins with the wrong clauses.

Therefore, in many cases it is better to use a calculus that starts with the query and works backward until the facts are reached. Such systems are designated backward chaining, in contrast to forward chaining systems, which start with facts and finally derive the query, as in the above example with the modus ponens. For backward chaining of Horn clauses, SLD resolution is used. This leads to a great reduction of the search space.

The literals of the negated query are the goals. This process continues until the list of subgoals of the current clauses the so-called goal stack is empty. With that, a contradiction has been found. If, for a subgoal ¬Bi , there is no clause with the complementary literal Bi as its clause head, the proof terminates and no contradiction can be found.

The query is thus unprovable. SLD resolution plays an important role in practice because programs in the logic programming language PROLOG consist of predicate logic Horn clauses, and their processing is achieved by means of SLD resolution see Exercise 2.

Thus the sets of unsatisfiable, satisfiable, and valid formulas are decid- able. The computation time of the truth table method for satisfiability grows in the worst case exponentially with the number n of variables because the truth table has 2n rows. An optimization, the method of semantic trees, avoids looking at variables that do not occur in clauses, and thus saves computation time in many cases, but in the worst case it is likewise exponential.

In resolution, in the worst case the number of derived clauses grows exponen- tially with the number of initial clauses. To decide between the two processes, we can therefore use the rule of thumb that in the case of many clauses with few vari- ables, the truth table method is preferable, and in the case of few clauses with many variables, resolution will probably finish faster. The question remains: can proof in propositional logic go faster? Are there better algorithms? The answer: probably not.

After all, S. Cook, the founder of complexity theory, has shown that the 3-SAT problem is NP-complete. For Horn clauses, however, there is an algorithm in which the computation time for testing satisfiability grows only linearly as the number of literals in the formula increases. For example, the verification of digital circuits and the gener- ation of test patterns for testing of microprocessors in fabrication are some of these tasks.

Special proof systems that work with binary decision diagrams BDD are also employed as a data structure for processing propositional logic formulas. In AI, propositional logic is employed in simple applications. For example, sim- ple expert systems can certainly work with propositional logic. However, the vari- ables must all be discrete, with only a few values, and there may not be any cross- relations between variables.

Complex logical connections can be expressed much more elegantly using predicate logic. Probabilistic logic is a very interesting and current combination of propositional logic and probabilistic computation that allows modeling of uncertain knowledge. It is handled thoroughly in Chap. Fuzzy logic, which allows infinitely many truth values, is also discussed in that chapter.

Exercise 2. To avoid a costly syntax check of the formulas, you may represent clauses as lists or sets of literals, and the formulas as lists or sets of clauses. The pro- gram should indicate whether the formula is unsatisfiable, satisfiable, or true, and output the number of different interpretations and models.

b Show that the resolution rule 2. Present the result in CNF. a The XOR operation exclusive or between two variables. b The statement at least two of the three variables A, B, C are true. The criminal had no accomplice and did not have the key, or he had the key and an accomplice. The criminal had the key. Did the criminal come in a car or not? b Exercise 2. First-order Predicate Logic 3 Many practical, relevant problems cannot be or can only very inconveniently be formulated in the language of propositional logic, as we can easily recognize in the following example.

Assume of these robots can stop anywhere on a grid of × points. The definition of relationships between objects here robots becomes truly difficult. In first-order predicate logic, we can define for this a predicate Position number, xPosition, yPosition. Ertel, Introduction to Artificial Intelligence, 31 Undergraduate Topics in Computer Science, DOI Definition 3. The sets V , K and F are pairwise disjoint.

Some examples of terms are f sin ln 3 , exp x and g g g x. To be able to establish logical relationships between terms, we build formulas from terms. Variables which are not in the scope of a quantifier are called free variables. In Table 3. In predicate logic, the meaning of formulas is recursively defined over the construction of the formula, in that we first assign constants, variables, and function symbols to objects in the real world.

Every n-place function symbol is assigned an n-place function. Every n-place predicate symbol is assigned an n-place relation. Example 3. The pair 1, 3 is not a member of G.

The formula F is false under the interpretation I2. Obviously, the truth of a formula in PL1 depends on the interpretation.

Now, after this preview, we define truth. The definitions of semantic equivalence of formulas, for the concepts satis- fiable, true, unsatisfiable, and model, along with semantic entailment Def- initions 2.

The edges going from Clyde B. upward to Mary B. and Oscar B. represent the element Clyde B. as a child relationship Theorem 3. is a child of Karen A. and Frank A. We now want to establish formulas for family relationships.

For further definitions we refer to Exercise 3. Now we build a small knowledge base with rules and facts. We can now ask, for example, whether the propositions child eve, oscar, anne or descendant eve, franz are derivable. To that end we require a calculus.

The equality of terms in mathematics is an equivalence relation, meaning it is reflex- ive, symmetric and transitive. If we want to use equality in formulas, we must either incorporate these three attributes as axioms in our knowledge base, or we must in- tegrate equality into the calculus. Often a variable must be replaced by a term. To carry this out correctly and describe it simply, we give the following definition.

Thereby we do not allow any variables in the term t that are quantified in ϕ. In those cases variables must be renamed to ensure this. Through this equivalence, universal, and existential quantifiers are mutually replace- able. However, they are disruptive for automatic inference in AI because they make the structure of formulas more complex and increase the number of applicable inference rules in every step of a proof.

Therefore our next goal is to find, for every predicate logic formula, an equivalent formula in a standardized normal form with as few quantifiers as possible.

As a first step we bring universal quantifiers to the beginning of the formula and thus define Definition 3. We see then that in 3. Rather, we must first eliminate the implications so that there are no negations on the quanti- fiers. It holds in general that we may only pull quantifiers out if negations only exist directly on atomic sub-formulas.

The transformed formula is equivalent to the output formula. The fact that this transformation is always possible is guaranteed by Theorem 3. In addition, we can eliminate all existential quantifiers. However, the formula resulting from the so-called Skolemization is no longer equivalent to the output for- mula.

Its satisfiability, however, remains unchanged. Because the variable y1 apparently depends on x1 and x2 , every occurrence of y1 is replaced by a Skolem function g x1 , x2. It is important that g is a new function symbol that has not yet appeared in the formula. Transformation into prenex normal form: Transformation into conjunctive normal form Theorem 2.

Elimination of implications. Renaming of variables if necessary. Factoring out universal quantifiers. Skolemization: Replacement of existentially quantified variables by new Skolem functions. Deletion of resulting universal quantifiers. The skolemized prenex and conjunctive normal form of 3.

By dropping the variable n0 , the Skolem function can receive the name n0. The procedure for transforming a formula in conjunctive normal form is sum- marized in the pseudocode represented in Fig. Skolemization has polynomial runtime in the number of literals. When transforming into normal form, the number of literals in the normal form can grow exponentially, which can lead to exponential computation time and exponential memory usage.

The reason for this is the repeated application of the distributive law. The actual problem, which results from a large number of clauses, is the combinatorial explosion of the search space for a sub- sequent resolution proof.

However, there is an optimized transformation algorithm which only spawns polynomially many literals [Ede91]. In the next section we will primarily concentrate on the resolution calculus, which is in practice the most important efficient, automatizable calculus for formulas in conjunctive normal form.

Here, using Example 3. When eliminating universal quantifiers one must keep in mind that the quantified variable x must be replaced by a ground term t, meaning a term that contains no variables. The proof of child eve, oscar, anne from an appropriately reduced knowledge base is presented in Table 3.

The two formulas of the reduced knowledge base are listed in rows 1 and 2. In row 3 the universal quantifiers from row 2 are eliminated, and in row 4 the claim is derived with modus ponens. The calculus consisting of the two given inference rules is not complete.

Please check back soon for future events, and sign up to receive invitations to our events and briefings. December 1, Speaker Series on California's Future — Virtual Event. November 30, Virtual Event. November 18, Annual Water Conference — In-Person and Online. We believe in the power of good information to build a brighter future for California.

Help support our mission. Mark Baldassare , Dean Bonner , Rachel Lawler , and Deja Thomas. Supported with funding from the Arjay and Frances F. Miller Foundation and the James Irvine Foundation. California voters have now received their mail ballots, and the November 8 general election has entered its final stage.

Amid rising prices and economic uncertainty—as well as deep partisan divisions over social and political issues—Californians are processing a great deal of information to help them choose state constitutional officers and state legislators and to make policy decisions about state propositions.

The midterm election also features a closely divided Congress, with the likelihood that a few races in California may determine which party controls the US House. These are among the key findings of a statewide survey on state and national issues conducted from October 14 to 23 by the Public Policy Institute of California:.

Today, there is a wide partisan divide: seven in ten Democrats are optimistic about the direction of the state, while 91 percent of Republicans and 59 percent of independents are pessimistic.

Californians are much more pessimistic about the direction of the country than they are about the direction of the state. Majorities across all demographic groups and partisan groups, as well as across regions, are pessimistic about the direction of the United States. A wide partisan divide exists: most Democrats and independents say their financial situation is about the same as a year ago, while solid majorities of Republicans say they are worse off. Regionally, about half in the San Francisco Bay Area and Los Angeles say they are about the same, while half in the Central Valley say they are worse off; residents elsewhere are divided between being worse off and the same.

The shares saying they are worse off decline as educational attainment increases. Strong majorities across partisan groups feel negatively, but Republicans and independents are much more likely than Democrats to say the economy is in poor shape. Today, majorities across partisan, demographic, and regional groups say they are following news about the gubernatorial election either very or fairly closely.

In the upcoming November 8 election, there will be seven state propositions for voters. Due to time constraints, our survey only asked about three ballot measures: Propositions 26, 27, and For each, we read the proposition number, ballot, and ballot label.

Two of the state ballot measures were also included in the September survey Propositions 27 and 30 , while Proposition 26 was not. This measure would allow in-person sports betting at racetracks and tribal casinos, requiring that racetracks and casinos offering sports betting make certain payments to the state to support state regulatory costs.

It also allows roulette and dice games at tribal casinos and adds a new way to enforce certain state gambling laws. Fewer than half of likely voters say the outcome of each of these state propositions is very important to them. Today, 21 percent of likely voters say the outcome of Prop 26 is very important, 31 percent say the outcome of Prop 27 is very important, and 42 percent say the outcome of Prop 30 is very important.

Today, when it comes to the importance of the outcome of Prop 26, one in four or fewer across partisan groups say it is very important to them. About one in three across partisan groups say the outcome of Prop 27 is very important to them. Fewer than half across partisan groups say the outcome of Prop 30 is very important to them. When asked how they would vote if the election for the US House of Representatives were held today, 56 percent of likely voters say they would vote for or lean toward the Democratic candidate, while 39 percent would vote for or lean toward the Republican candidate.

Democratic candidates are preferred by a point margin in Democratic-held districts, while Republican candidates are preferred by a point margin in Republican-held districts.

Abortion is another prominent issue in this election. When asked about the importance of abortion rights, 61 percent of likely voters say the issue is very important in determining their vote for Congress and another 20 percent say it is somewhat important; just 17 percent say it is not too or not at all important.

With the controlling party in Congress hanging in the balance, 51 percent of likely voters say they are extremely or very enthusiastic about voting for Congress this year; another 29 percent are somewhat enthusiastic while 19 percent are either not too or not at all enthusiastic. Today, Democrats and Republicans have about equal levels of enthusiasm, while independents are much less likely to be extremely or very enthusiastic. As Californians prepare to vote in the upcoming midterm election, fewer than half of adults and likely voters are satisfied with the way democracy is working in the United States—and few are very satisfied.

Satisfaction was higher in our February survey when 53 percent of adults and 48 percent of likely voters were satisfied with democracy in America. Today, half of Democrats and about four in ten independents are satisfied, compared to about one in five Republicans. Notably, four in ten Republicans are not at all satisfied.

In addition to the lack of satisfaction with the way democracy is working, Californians are divided about whether Americans of different political positions can still come together and work out their differences.

Forty-nine percent are optimistic, while 46 percent are pessimistic. Today, in a rare moment of bipartisan agreement, about four in ten Democrats, Republicans, and independents are optimistic that Americans of different political views will be able to come together. Notably, in , half or more across parties, regions, and demographic groups were optimistic.

Today, about eight in ten Democrats—compared to about half of independents and about one in ten Republicans—approve of Governor Newsom. Across demographic groups, about half or more approve of how Governor Newsom is handling his job. Approval of Congress among adults has been below 40 percent for all of after seeing a brief run above 40 percent for all of Democrats are far more likely than Republicans to approve of Congress.

Fewer than half across regions and demographic groups approve of Congress. Approval in March was at 44 percent for adults and 39 percent for likely voters. Across demographic groups, about half or more approve among women, younger adults, African Americans, Asian Americans, and Latinos. Views are similar across education and income groups, with just fewer than half approving. Approval in March was at 41 percent for adults and 36 percent for likely voters.

Across regions, approval reaches a majority only in the San Francisco Bay Area. Across demographic groups, approval reaches a majority only among African Americans.

This map highlights the five geographic regions for which we present results; these regions account for approximately 90 percent of the state population. Residents of other geographic areas in gray are included in the results reported for all adults, registered voters, and likely voters, but sample sizes for these less-populous areas are not large enough to report separately.

The PPIC Statewide Survey is directed by Mark Baldassare, president and CEO and survey director at the Public Policy Institute of California. Coauthors of this report include survey analyst Deja Thomas, who was the project manager for this survey; associate survey director and research fellow Dean Bonner; and survey analyst Rachel Lawler. The Californians and Their Government survey is supported with funding from the Arjay and Frances F.

Findings in this report are based on a survey of 1, California adult residents, including 1, interviewed on cell phones and interviewed on landline telephones. The sample included respondents reached by calling back respondents who had previously completed an interview in PPIC Statewide Surveys in the last six months.

Interviews took an average of 19 minutes to complete. Interviewing took place on weekend days and weekday nights from October 14—23, Cell phone interviews were conducted using a computer-generated random sample of cell phone numbers. Additionally, we utilized a registration-based sample RBS of cell phone numbers for adults who are registered to vote in California. All cell phone numbers with California area codes were eligible for selection.

After a cell phone user was reached, the interviewer verified that this person was age 18 or older, a resident of California, and in a safe place to continue the survey e. Cell phone respondents were offered a small reimbursement to help defray the cost of the call.

Cell phone interviews were conducted with adults who have cell phone service only and with those who have both cell phone and landline service in the household. Landline interviews were conducted using a computer-generated random sample of telephone numbers that ensured that both listed and unlisted numbers were called. Additionally, we utilized a registration-based sample RBS of landline phone numbers for adults who are registered to vote in California. All landline telephone exchanges in California were eligible for selection.

For both cell phones and landlines, telephone numbers were called as many as eight times. When no contact with an individual was made, calls to a number were limited to six. Also, to increase our ability to interview Asian American adults, we made up to three additional calls to phone numbers estimated by Survey Sampling International as likely to be associated with Asian American individuals.

Accent on Languages, Inc. The survey sample was closely comparable to the ACS figures. To estimate landline and cell phone service in California, Abt Associates used state-level estimates released by the National Center for Health Statistics—which used data from the National Health Interview Survey NHIS and the ACS. The estimates for California were then compared against landline and cell phone service reported in this survey. We also used voter registration data from the California Secretary of State to compare the party registration of registered voters in our sample to party registration statewide.

The sampling error, taking design effects from weighting into consideration, is ±3. This means that 95 times out of , the results will be within 3. The sampling error for unweighted subgroups is larger: for the 1, registered voters, the sampling error is ±4. For the sampling errors of additional subgroups, please see the table at the end of this section. Sampling error is only one type of error to which surveys are subject. Results may also be affected by factors such as question wording, question order, and survey timing.

We present results for five geographic regions, accounting for approximately 90 percent of the state population. Residents of other geographic areas are included in the results reported for all adults, registered voters, and likely voters, but sample sizes for these less-populous areas are not large enough to report separately. We also present results for congressional districts currently held by Democrats or Republicans, based on residential zip code and party of the local US House member.

We compare the opinions of those who report they are registered Democrats, registered Republicans, and no party preference or decline-to-state or independent voters; the results for those who say they are registered to vote in other parties are not large enough for separate analysis. We also analyze the responses of likely voters—so designated per their responses to survey questions about voter registration, previous election participation, intentions to vote this year, attention to election news, and current interest in politics.

The percentages presented in the report tables and in the questionnaire may not add to due to rounding. Additional details about our methodology can be found at www. pdf and are available upon request through surveys ppic. October 14—23, 1, California adult residents; 1, California likely voters English, Spanish.

Margin of error ±3.

He developed computer games in which the players, distributed over the network, are supposed to collaboratively describe pictures with key words. b Exercise 2. As a basic description framework, the World Wide Web Consortium developed the language RDF Resource Description Framework. See comments. The query is thus unprovable. With the actuators it carries out actions and changes its environment. Here we observe that append is not a two-place function, but a three-place relationship.

Categories:

• Binary Options
• Forex Brokers
• Forex
• Forex Bitcoin
• Binary Options online