Semantic search bot

Slack/Telegram bot with semantic search by knowledge base

The task of the project:
Making the process of searching for textual information easier
What we did and how we did it:
Basing on LSI model (we compared LDA, LCI, word2vex, and fastText models), we realized a prototype of a semantic search system by unstructured data (knowledge base from Slack).
-We prepared a knowledge base
-We conducted stemming, lemmatization, defined stop-words (forbidden words)
-We set the search quality assessment criterion (chose the expert assessment)
-We realized a Telegram-bot
We realized a mobile application for iOS
Time-frame: 2 months

Project team
DEV: Alexandra Yakimenko, Ekaterina Simonova
RnD: Evgeny Khomutov, Vadim Kondaratsev, Alexander Korneev
PM: Oleg Yusupov

Link to the bot: @phyge bot
Link to the git

Year of the project implementation: 2018