RuPersonaAgent

.

The purpose of the project

The rupersonaagent library is designed for quick and easy development of a Russian-language personalized conversational agent with dynamic long-term memory. Each method and model can also be used separately for purposes other than developing a conversational agent. The library works with python3.8

Table of Contents

• Core features
• Installation
• Examples
• Project Structure
• Documentation
• License
• Acknowledgments
• Contacts
• Citation

Core features

• Speech Extraction
• Generative Model
• Long-term and Internet memory
• Persona Knowledge Extraction
• Rule Based Information Extraction
• Inference Optimization

Installation

RuPersonaAgent can be installed with pip:

  $ pip install rupersonaagent

Examples

link for usage examples

Project Structure

Stable version is located in the main branch

Repo includes:

• generative_model - methods for creating and working with generative Russian language models
• speech_extraction - methods for extracting the main characteristics of speech from datasets or individual cues, a method for assessing the similarity of speech characteristics
• rule_based_information_extraction - methods for extracting important facts about a person from a text based on linguistic rules

Documentation

Documentation

License

License

Acknowledgments

Gratitude to ITMO University for sponsoring and supporting this project

Contacts

Telegram channel Q&A RuPersonaAgent

Citation

Матвеева А.А., Махныткина О.В. Метод аугментации текстовых данных с сохранением стиля речи и лексики персоны // Научно-технический вестник информационных технологий, механики и оптики. 2023. Т. 23, № 4. С. 743–749. doi: 10.17586/2226-1494-2023-23-4-743-749

Посохов П.А., Рудалева Е.А., Скрыльников С.С., Махныткина О.В., Кабаров В.И. Извлечение информации о персоне из диалоговых данных на русском языке //Информационные технологии. - 2023. в печати