G. B. Bronfeld – Ph.D. (Eng.), Associate Professor, Nizhniy Novgorod State Technical University n.a. R.E. Alekseev
The article first provides a brief analysis of approaches to the simulation of knowledge in the texts in a natural language (NL). Explored framework of perception of knowledge by the specialists from the texts, as well as, those that will input knowledge from them in the knowledge database (KB) created by intelligent systems (IS), in particular, expert system (ES). Shown, that their understanding of the content in the text differs considerably. One of the fundamental problems of many models of knowledge used for KB IS, is that problems arise when replying in a linked text which can be understood by the expert.
Proposed is a model of knowledge in the texts that allows at once to prepare certain blocks, which can be used to output the results and that, no less important, will fundamentally reduce the amount of generated KB compared to some other approaches. It is known from linguistics that «simple sentence ... has been and still remains a basic unit of syntax of the text» and «the most significant feature of a sentence is its ability to form and express a though». Therefore, the basis of the models of knowledge from the start is taken from a simple sentence, built on grammar NL.
Every sentence of the text of an expert-editor (creating KB) presents a small semantic network with creation of thesaurus terms (including synonyms), relationship, qualitative and quantitative features and more.
The molinga have the form
D; R; Z; K; F; Q.
Here D is the set of identifiers with which the molingas stand out among the many molingas. Provided are all identifiers of the molinga of all texts, where the same knowledge was met. Similarity of meaning of sentences and receiving of the same molingas determines the expert-editor. Due to the identifiers in the response, it can recover close to the original text. R – is the condition of applicability of the kernel molinga. The main element of molinga is the core of Z – simulated proposal (if you saved the proposals of any complexity). To specify a code sequence of dictionaries fixing position in the core of molinga – words that act as terminology, relationships, qualitative and quantitative attributes, and linguistic ties. In F are specified levels of reliability molingas in the form certainty factor. Q describes postconditions of molinga. They are updated when the core of molinga is realized.
The expert-editor consistently manually (semi-automatic) looking at the entire text of the Ti and turns the complex sentences of text to simple sentences and the last in the molinga, if necessary, adding their own. The molinga refers to the model views pronounced of declarative type.
The modeling process text is discussed in step-by-step. Provided an example of the representation of the text by molinga with compiling of dictionaries. Described are the different ways of formal modeling of parts of the text headers, formula-based calculations, lists, tables.
This method of modeling knowledge can be used for new information systems, in particular, elingas, and modernization of the existing ones, for example, knowledge management. It allows for realization of the idea of integration of knowledge, developed once a professor E. Tyugu, the establishment of the united KB-based blend of knowledge tested in the late 90's by J. Gray, and implementing ideas granulation of knowledge, but in a natural way.
Gavrilova T.A., Khoroshevskiy V.F. Bazy znaniy
intellektual'nykh sistem. SPb.: Piter. 2001. 384 s.
Bargasegyan A.A., Kupriyanov M.S., Stepanenko
V.V., Kholod I.I. Tekhnologii analiza dannykh: Data Mining, Visual Mining, Text
Mining, OLAP. SPb.: BKhV-Peterburg. 2008. 384 s.
Popper K. Logika i rost nauchnogo znaniya.
Izb. raboty. M.: Progress. 1983. 302 s.
Khaykin S. Neyronnye seti: polnyy kurs. M.:
Vil'yams. 2008. 1104 s.
Tsouricov et al. Document semantic analysis /selection with knowledge
creativity capability utilizing subject-action-object (SAO) structures. 2000.
Sovpel' I.V. Sistema avtomaticheskogo izvlecheniya
znaniy iz teksta i ee prilozheniya // Iskusstvennyy intellekt. 2004. № 3. S. 668–679.
Y.A. Grammar, Meaning and the Machine Analysis of language. London: Londons
Routledge and Kegan Paul. 1972. 94 p.
Popov E.V. Obshchenie s EVM
na estestvennom yazyke. M.: Nauka. 1982. 360 s.
Mel'chuk I.A. Opyt teorii
lingvisticheskikh modeley «Smysl–Tekst».Semantika, sintaksis. M.: Shkola
«Yazyki russkoy kul'tury». 1999. 346 s.
Vorozhtsov A.V. Industriya
znaniy // Informatsionnye tekhnologii i vychislitel'nye sistemy. 2003. № 4.
Bronfel'd G.B. Ob odnom
podkhode k postroeniyu intellektual'nykh elektronnykh knig i o molinge // Trudy
VIII Mezhdunar. simpoziuma «Intellektual'nye sistemy». M.: RUSAKI. 2008. S.
Bronfel'd G.B. Podkhod k
integratsii znaniy s pomoshch'yu eling // Trudy NGTU im. R.Ye. Alekseeva
«Sistemy obrabotki informatsii i upravleniya». 2009. T.74. Vyp. 15. S. 5–11.
Bronfel'd G.B. Pryamoe
nalozhenie znaniy i nekotorye ego osobennosti // Trudy ros. konf.
«Tekhnicheskie i programmnye sredstva sistem upravleniya, kontrolya i
izmereniya (UKI-10)». M.: IPU im. V.A. Trapeznikova RAN. 2010. S. 201–211.