Abstract—This paper describes the design of a knowledge representation and reasoning system, named Dolphin, which is based on higher-order temporal Transparent Intensional Logic (TIL). An intelligent agent (NAM), that is able to read newspaper headlines from specialized internet server and allows users to ask questions about various world situations is chosen to demonstrate Dolphin features. Temporal aspects play an essential role in natural language therefore we present how this phenomenon is handled in the system. Reasoning capabilities of the agent are divided into three individual strategies and described in the text. As a result we compare NAM answers to one of the most used search engines nowadays. Index Terms—Transparent Intensional Logic, Knowledge Base, Dolphin, Inference, Temporal aspect, News answering machine I. INTRODUCTION Today's web searching tools are mainly based on full-text search methods which detect key words in the input text and then look them up by fast database algorithms [6]. Such approach is efficient in speed, but it inevitably ignores time information and natural form of human communication. The Dolphin knowledge base system that is presented in this paper is based on higher-order typed Transparent Intensional Logic (TIL) that is capable of analyzing natural language (NL) and representing its meaning in an algorithmically accessible form. In contrast to other logic systems, TIL is designed to properly analyze time, personal attitudes and belief sentences which makes it a perfect tool for NL sentences meaning representation and its algorithmic semantic analysis [1]. Dolphin uses TIL as a bridge between human language and a computer. Early stages of the project can be found in [2], where we present the first architecture design and concept. The current analysis uses more complex and powerful approaches but the original ideas still provide good learning points about Dolphin. For the presentation of actual current features of the system, we have chosen a real application based on daily news – News answering machine (NAM). Fig. 1 displays basic parts of NAM. The news are represented by a dedicated web server that holds a set of daily Manuscript received May 24, 2012; revised July 27, 2012. This work has been partly supported by the Ministry of Education of CR within the Center of basic research LC536 and by the Czech Science Foundation under the project P401/10/0792. The authors are with Masaryk University, Faculty of Informatics, Czech republic (e-mail: xgardon@fi.muni.cz; xhorak@fi.muni.cz). news – as far as the system is not yet capable of analyzing all natural language phenomena, only certain subset of NL sentences is allowed. NLD is a module, which translates news from the form of NL text to a meaning representation language readable by a computer. The language is defined by the theory of TIL and is implemented as text-based DOLLY language that allows to effectively process lambda-calculus entities [3, p. 6-12]. An example of a NL input, it’s TIL translation and the corresponding DOLLY script follows: Apple is red. x…ι: λwλt [ red wt x ] [ Apple wt x ] {True/o := \w\t [[[ And [[[red w_Dolphin] time] x]] […]]} The translation mechanism from NL sentences to DOLLY is modular. The currently tested implementation is based on SYNT [4]. SYNT takes a NL sentence in the Czech or English language and thanks to syntax, semantic and corpus processing; it produces the DOLLY transcription(s) of the sentence. The Dolphin NAM itself does not contain any other language specific limitation than those impacted by the selected NL-to-DOLLY translation module. Fig. 1. Basic parts of News answering machine (NAM). II. THE DOLPHIN INTERNALS TIL works with objects of universe and their constructions instead of words. As you can see in Fig. 2A, a word depicts a construction which represents the meaning procedure pointing to the referent object. The sentence itself is, of course, also analyzed with a construction constructing a proposition (possible world and time dependent truth value, see Fig. 2B). Treating the (structural) meaning as constructions (and sub-constructions) is than clear and allows us to connect attributes and methods with the meaning objects or their algorithmic form called Dolly Construction (DC). In this way a sentence “Space is unlimited” is stored as three interconnected DCs. A question may arise why three DCs are used as only two words (Space, unlimited) are present. Other interpretation of the sentence “It is true that Space is unlimited” clearly shows the answer – sentence as a whole constructs the (intensional) Knowledge Base for Transparent Intensional Logic and Its Use in Automated Daily News Retrieval and Answering Machine A. Gardoň and A. Horák International Journal of Machine Learning and Computing, Vol. 2, No. 4, August 2012 487