• A
  • A
  • A
  • ABC
  • ABC
  • ABC
  • А
  • А
  • А
  • А
  • А
Regular version of the site
For visually-impairedUser profile (HSE staff only)SearchMenu

Development and Research of New Methods for Constructing Distributed Power-Saving Computing Systems for Semantic Search of Text Information Based on Dataflow Computation Model

Head: Leokhin, Yuriy

Object of research: natural language (NL) semantic analysis system based on a dataflow distributed computing system

Goal of research: development of a complete concept covering all stages and levels of the development of the NL semantic analysis computing system

Research tasks:

1. Analysis of the research scope and the choice of further direction for research

2. Development of the NL semantic analysis technique, ensuring effective functioning with lexemes and language structure polysemanticism

3. Data format development for language description in the NL semantic analysis system

4. Development of a method to describe the NL processing rules

5. Development of the formal model of the NL semantic analysis computing system

6. Development of the distributed computing system architecture (list of devices that are a part of the computing system; sets of devices commands; function charts, etc.) for NL semantic analysis and semantic search

7. Development of a programming language for the knowledge base description

8. Construction of an experimental programme model of the NL semantic analysis computing system

The novelty of the research work is that the dataflow architecture is chosen for realization of NL analysis computing system (CS). To be exact, the object-attribute CS architecture is a proprietary development of HSE MIEM. The novelty of the obtained solution is also the fact that the proposed NL semantic analysis principles cover all stages and levels of the CS, such as the hardware and software architecture levels, programming language, data format, and formalization of the NL semantic analysis rules. All levels work on the basis of a uniform philosophy of computing process organization.

Results of research:

- data format for lexeme description

- principles of the NL semantic analysis

- distributed CS architecture for the NL semantic analysis

- formal model for description of the text semantic net synthesis rules

- formal model for the description of the computing process in CS

- implementation of the experimental knowledge base for the Russian and English languages analysis

Level of implementation: an experimental knowledge base for analysis of Russian and English languages has been developed

Field of application: information search on the internet, automatic text review, document classification, artificial intelligence systems, robotics, intelligent interfaces between people and computers


Леохин Ю. Л., Мягков А. С. Разработка подхода к организации потоков данных в многопроцессорной вычислительной системе с кольцевой архитектурой // Информатизация образования и науки. 2013. Т. 40. № 4. C. 58-67. 
Салибекян С. М., Панфилов П. Б. Формализация dataflow-модели вычислительного процесса, in: Объектные системы - 2013: материалы VII Международной научно-практической конференции, Ростов-на-Дону, 10-12 мая 2013 г.. Ростов-на-Дону , 2013. С. 87-93. 
Leokhin, Y., Myagkov, A., Panfilov, P. Gomapreduce parallel computing model implementation on a cluster of plan9 virtual machines, in: 26th DAAAM International Symposium on Intelligent Manufacturing and Automation 2015. New York : Curran Associates, Inc., 2015. С. 0656 - 0662.