Voice-Interactive Semantic Search Interface with Vector Databases

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F49777513%3A23520%2F24%3A43973057" target="_blank" >RIV/49777513:23520/24:43973057 - isvavai.cz</a>

Výsledek na webu

<a href="https://svk.fav.zcu.cz/download/proceedings_svk_2024.pdf" target="_blank" >https://svk.fav.zcu.cz/download/proceedings_svk_2024.pdf</a>

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Voice-Interactive Semantic Search Interface with Vector Databases

Popis výsledku v původním jazyce

Semantic searching offers significant advantages over full-text search, particularly be- cause it allows users to formulate queries in natural language without needing to know the precise indexed key phrases. By using vector databases that store and index data as high- dimensional vectors, we can search through large datasets in real-time. In this work, we present a custom web-based interface for state-of-the-art semantic search on arbitrary textual data. Additionally, we integrate our in-house speech technologies - ASR and TTS to enhance user interaction. The interface supports two modes: 1) Searching based on retrieval- augmented generation (RAG) with an LLM generating answers in a chat-like format, and 2) raw semantic matching with indexed data. In both modes, the original PDF file is shown and the exact source of the retrieved information is provided.

Název v anglickém jazyce

Voice-Interactive Semantic Search Interface with Vector Databases

Popis výsledku anglicky

Semantic searching offers significant advantages over full-text search, particularly be- cause it allows users to formulate queries in natural language without needing to know the precise indexed key phrases. By using vector databases that store and index data as high- dimensional vectors, we can search through large datasets in real-time. In this work, we present a custom web-based interface for state-of-the-art semantic search on arbitrary textual data. Additionally, we integrate our in-house speech technologies - ASR and TTS to enhance user interaction. The interface supports two modes: 1) Searching based on retrieval- augmented generation (RAG) with an LLM generating answers in a chat-like format, and 2) raw semantic matching with indexed data. In both modes, the original PDF file is shown and the exact source of the retrieved information is provided.

Druh

O - Ostatní výsledky

CEP obor

—

OECD FORD obor

20205 - Automation and control systems

Projekt

—

Návaznosti

S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2024

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů