OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Search & Research AI

Bertsearch

Bertsearch is a Python search application that uses a pretrained BERT model to vectorize document text, stores the vectors in Elasticsearch, and provides a browser interface for searching indexed documents.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Bertsearch — OpenViST project profile in Search & Research AI.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when you want a self-hosted, Docker-based starting point for semantic search over documents and can configure Docker, Elasticsearch, and a pretrained BERT model.

Who it is for

  • Developers building a semantic search prototype over their own text records.
  • Teams exploring job search over indexed titles and descriptions.

Documented uses

  • Create an Elasticsearch index with text and dense-vector fields.
  • Convert CSV document text into BERT vectors and index the resulting records.
  • Search the indexed documents through the browser interface.
Before you choose

Requirements & limits

The repository is public and licensed under MIT. The documented setup runs locally with Docker Compose and exposes a browser interface at 127.0.0.1:5000.

  • Requires Docker and Docker Compose 1.22.0 or later.
  • A pretrained BERT model must be downloaded and configured.
  • The Elasticsearch dense-vector dimensions must match the selected model.
  • The README recommends more than 8 GB of Docker memory where possible.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Set up Bertsearch and index CSV documents

Prepare the documented Docker-based search system and make CSV records available in its Elasticsearch index.

Before you start

  • Docker
  • Docker Compose 1.22.0 or later
  • A pretrained BERT model
  • CSV records with document text and optional title fields

Steps

  1. Download and unzip one of the pretrained BERT models listed in the README.

  2. Set PATH_MODEL to the model directory and INDEX_NAME to the Elasticsearch index name.

  3. Start the containers with docker-compose up; the README recommends assigning Docker more than 8 GB of memory where possible.

  4. Create the index with the example script, ensuring the text_vector dimensions match the selected BERT model.

  5. Use the document-creation and indexing scripts to convert the CSV records into vectorized documents and add them to the index.

Expected outcome

The records are indexed in Elasticsearch with BERT-generated vectors; the README identifies http://127.0.0.1:5000 as the browser interface.

Guide limitations

  • The vector field dimensions must match the selected pretrained model.
  • The README cautions that the BERT container needs high memory and recommends more than 8 GB for Docker where possible.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
16 October 2026
Original website title
Hironsan/bertsearch · Official homepage title · retained in its source language
How we review sourcesSuggest a correction