Journal influence
Higher Attestation Commission (VAK) - К1 quartile
Russian Science Citation Index (RSCI)
Bookmark
Next issue
№4
Publication date:
16 December 2025
Probabilistic inference in weakly formalized knowledge bases
Date of submission article: 01.03.2016
UDC: 004.89
The article was published in issue no. № 3, 2016 [ pp. 10-14 ]Abstract:The article introduces the process of probabilistic inference in weakly formalized knowledge bases. Semantic network is chosen as a graphical model of knowledge representation due to a convenient representation of automatically extracted data as a graph with links. The paper also contains a comparison of widely used production model approach with the proposed one. The authors describe main disadvantages of a production model approach that should be considered for developing such knowledge extraction systems. The aim of the work is new knowledge extraction from an automatically built knowledge base. Usually logical inference is used to achieve this aim in graphical models. In our case a domain model as well as a process of knowledge building (in particular by automatic or semi-automatic methods) restrict logical inference mechanism, so this algorithm is forced to work in conditions of uncertainty. Thus, standard logical inference algorithms provided for such model become irrelevant. The article proposes using probabilistic inference for the task and consequently using probabilistic inference programming language. The paper contains a comparison of several modern probabilistic logic programming languages like PRISM, ICL and ProbLog. The authors select a probabilistic logic programming language based on the results of this comparison. To implement probabilistic inference in a weakly formalized knowledge base we have selected ProbLog language (ProbLog2 in particular) that is a probability extension of Prolog.
Аннотация:В статье рассматривается процесс вероятностного вывода в слабоформализованных базах знаний. В качестве такой базы выбрана графическая модель представления знаний – семантическая сеть. Выбор обусловлен удобством представления автоматически извлеченных данных в виде графа со связями, а также удобством дальнейшего использования (чтения, изменения и поиска ошибок) данного графа. Также проводится сравнение широко используемого на данный момент продукционного подхода с предложенным, указываются основные недостатки продукционного подхода, которые необходимо учитывать при разработке подобных систем извлечения знаний. Целью исследования является извлечение новых знаний из автоматически полученных данных. Для достижения этой цели на графических моделях обычно производится логический вывод. Поскольку модель, а также способ получения данных (в данном случае автоматически или полуавтоматически) накладывают ограничения на механизм вывода, алгоритм вынужден работать в условиях неопределенности. Отсюда следует, что стандартные механизмы логического вывода, предусмотренные для данной модели, становятся неактуальными. В статье предлагается использовать вероятностный вывод и, следовательно, вероятностный язык логического программирования для его реализации. Также делается сравнение нескольких существующих языков вероятностного логического программирования, таких как PRISM, ICL и ProbLog. По результатам сравнения производится выбор языка вероятностного программирования для осуществления вывода. В качестве такого языка выбран язык ProbLog (в частности система ProbLog2), являющийся вероятностным расширением языка Prolog.
| Authors: Poleschuk E.A. (eapoleschuk@corp.ifmo.ru) - The National Research University of Information Technologies, Mechanics and Optics, St. Petersburg, Russia, Platonov A.V. (avplatonov@corp.ifmo.ru) - The National Research University of Information Technologies, Mechanics and Optics, St. Petersburg, Russia | |
| Keywords: probabilistic inference, semantic network, weakly formalized knowledge bases, problog, probabilistic logic programming, data uncertainty, relationships uncertainty |
|
| Page views: 14337 |
Print version Full issue in PDF (6.81Mb) Download the cover in PDF (0.36Мб) |
Использование вероятностного вывода в слабоформализованных базах знаний
DOI: 10.15827/0236-235X.115.010-014
Date of submission article: 01.03.2016
UDC: 004.89
The article was published in issue no. № 3, 2016. [ pp. 10-14 ]
The article introduces the process of probabilistic inference in weakly formalized knowledge bases. Semantic network is chosen as a graphical model of knowledge representation due to a convenient representation of automatically extracted data as a graph with links. The paper also contains a comparison of widely used production model approach with the proposed one. The authors describe main disadvantages of a production model approach that should be considered for developing such knowledge extraction systems.
The aim of the work is new knowledge extraction from an automatically built knowledge base. Usually logical inference is used to achieve this aim in graphical models. In our case a domain model as well as a process of knowledge building (in particular by automatic or semi-automatic methods) restrict logical inference mechanism, so this algorithm is forced to work in conditions of uncertainty. Thus, standard logical inference algorithms provided for such model become irrelevant.
The article proposes using probabilistic inference for the task and consequently using probabilistic inference programming language. The paper contains a comparison of several modern probabilistic logic programming languages like PRISM, ICL and ProbLog. The authors select a probabilistic logic programming language based on the results of this comparison. To implement probabilistic inference in a weakly formalized knowledge base we have selected ProbLog language (ProbLog2 in particular) that is a probability extension of Prolog.
Poleschuk E.A. (eapoleschuk@corp.ifmo.ru) - The National Research University of Information Technologies, Mechanics and Optics, St. Petersburg, Russia, Platonov A.V. (avplatonov@corp.ifmo.ru) - The National Research University of Information Technologies, Mechanics and Optics, St. Petersburg, Russia
Ссылка скопирована!
| Permanent link: http://swsys.ru/index.php?page=article&id=4171&lang=en |
Print version Full issue in PDF (6.81Mb) Download the cover in PDF (0.36Мб) |
| The article was published in issue no. № 3, 2016 [ pp. 10-14 ] |
The article was published in issue no. № 3, 2016. [ pp. 10-14 ]
Perhaps, you might be interested in the following articles of similar topics:Perhaps, you might be interested in the following articles of similar topics:
