Reviews of Some of the Top Academic Journals in Theoretical Computer Science and Artificial Intelligence
Artificial Intelligence is the oldest academic journal of the three and began in 1970. It contains the result of various researches in the AI field. This is a general AI journal and includes topics from an extremely broad range. It welcomes both papers on theoretical aspects of AI and applied versions. Specific examples of topics accepted into AI include: automated reasoning, computation theories of learning, heuristic search, knowledge representation, qualitative physics, signal, image and speech understanding, robotics, natural language understanding, and software and hardware architectures for AI. This journal reports on both results achieved and proposals for new ways of looking at problems in AI. With its broad range it has the largest intended audience of the three journals, which includes AI specialists, Computer Scientists, Cognitive Scientists, Psychologists, and Linguists. This journal would be an excellent beginning for any AI research.
Applied Artificial Intelligence mainly contains articles on the application aspect of AI. In addition to this it also acts as an exchange for ideas and thoughts about AI research and the impact of AI systems for solving various real world tasks in engineering, administration, and education. Articles may evaluate existing AI systems and even contain information or analysis on the cultural and social impacts of these AI systems. The intended audience is researchers, but they may be involved in any aspect of artificial intelligence to include robotics, computer information science, or hardware/software vendors.
Theoretical Computer Science is the most highly specialized journal with all its articles relating to theoretical structures which usually involve high mathematics and are very abstract. This journal does not address applications to AI specifically, but instead contains articles on the modeling and theoretical approach to methods of automata and neural computing which would/could be implemented in modern AI systems. The information theory could be building blocks in creating cutting edge AI which may or may not have real world implementations as of yet. Its intended audience is only researchers in the theoretical aspects of computer science.
The journal articles of Theoretical Computer Science are in the classical IMRAD style and contain original research and/or discussion of new approaches to classical problems in theoretical computer science. A very high level of mathematical understanding is required even for basic understanding of most of the material. No term definitions or expression expansions are given because of the assumed extreme level of technically competence of the intended audience. Examples of terms used without the definition given include distributed parallel computing models, membrane computing models, P systems, lattice isomorphism’s, and semi-groups. A full understanding of propositional logic, graph, and set theoretic is a prerequisite for even basic understanding. Full Mathematical proofs are given for completeness, but lack detailed explanation or analysis.
Most of the journal articles include detailed comparisons and contrasts and the results are kept in the theoretical realm with very little discussion of their applications in engineering or programming AI. This keeps this journal very “pure” in its approach to solving or furthering the solutions of high level problems. The graphics and visuals are kept to an absolute minimum, but are utilized to illustrate modeling systems and other ideas that may be ambiguous or harder to understand in only written words. The text is plain and there are no visual or graphic aids to improve reading or skimming the material. The articles contained in this journal are on average around 24 pages in length and include a very hefty source list. The abstracts average about a page and contain information on the history of the problem being discussed.
The content can be broke up into three distinct categories which each cover one extremely specific aspect of theoretical computer science: ‘Algorithms, automata, complexity and games’, ‘Logic, semantics and theory of programming’, and ‘Natural Computing’. Each category contains a host of research articles utilizing field specific terms without giving any definitions or further expansions (high specific prior knowledge is still assumed). The content is broken up into these even more specific categories for optimum efficiency and organization for the specific intended audience. All articles are primary literature and have not been published elsewhere. ‘Algorithms, automa, complexity and games’ is devoted to the study of algorithms and their complexity using analytical, combinatorial or probabilistic methods. ‘Logic, semantics and theory of programming’, is devoted to formal methods to check properties of programs or implement formally described languages. “Natural computing, is devoted to the study of computing occurring in nature and computing inspired by nature which include Evolutionary computing neural networks, molecular computing, and quantum computing.
All the content topics further the theoretical understanding of when and why things work. This is opposition of applied AI or engineering which is interesting in making things work and solving real world applications of concepts and technology. Many of these ideas may be utilized in new artificial intelligence technology, which is an even newer field and is still in rudimentary development, with many of it’s ideas and or theory’s just now coming into practice (It was only a short time ago a computer could not beat the human chess champion.)
In addition this publication produces a large number of special editions that cover materials and/or works presented at field related colloquiums. For example a recent special edition was devoted to the Third International Colloquium on words, Languages and Combinatories, which was held at Kyoto Sangyo University and consisted of selection of the papers presented there.
The methods and materials sections involve mostly modeling of systems in a new or altered way and applying logic and mathematics to these new models. In some cases computer simulations are performed and applications to test the theoretical models are undergone, but in most of the journals entries everything is left in the abstract with math and logic being utilized on an abstract object. A large amount of examples are included and are mostly in the form of a mathematical or abstract representation of the problem being brought into question. This allows for “thought” type experiments and challenges to existing norms in the fattest manner possible without the researcher even having to wait for real world implementations of the discussed systems of computing.
This journal would be useful for computer science researchers and/or students with an extremely high technical expertise interested in different ways or methods of approaching a complex problem. This problem should be in an abstract form and not require an immediate or timely application. The scope of the journal is limited to complex computer science problems in the theoretical sense (hence the name Theoretical Computer Science) and does not include any details about real world applications or implementation, thus rendering it usefully only to individuals, researchers, peers etc interested in ways of solving and approaching complex computer science problems in the ideal or theoretical sense and gaining information on the types and limitations of current computing systems.
