Skip to content

Exploratory search engine based on hierarchical topic models from BigARTM

Notifications You must be signed in to change notification settings


Repository files navigation


Rysearch screenshots Rysearch is an exploratory search engine and recommender system. Based on MongoDB and BigARTM, it allows to perform both exact and inexact search queries over popular-scientific corpora and visualizes these corpora in a hierarchical "map of knowledge", which is built using weakly supervised hierarchical topic models.

The demonstration of the current stable version can be found here.

How to run Rysearch?

The preferred way to install and run Rysearch is via Docker. You can either pull the latest containers from Docker hub or build everything on your own. Previously, Rysearch could also be built using Nix; this is now deprecated, but the corresponding .nix files are retained for the reference.


Step 1: Obtaining Docker containers

The easiest way to get the containers is to pull them from Docker hub:

git clone /path/to/Rysearch
cd /path/to/Rysearch/docker
docker-compose pull

Alternatively, it is possible to build the required containers on your own infrastructure:

git clone /path/to/Rysearch
cd /path/to/Rysearch/docker
docker-compose build

Step 2: Running the containers

After the containers are either downloaded or manually built, you can use docker-compose to run them:

cd /path/to/Rysearch/docker
docker-compose up

By default, docker-compose runs a single worker to process all search queries. You can run an arbitrary number of workers, say N workers, to balance the load, like this:

cd /path/to/Rysearch/docker
docker-compose up --scale bridge=N


If you are planning to use Rysearch in your research projects, please cite one of the following articles:

  • Anton Belyy. Construction and quality evaluation of heterogeneous hierarchical topic models. Bachelor's thesis, 2018. [thesis] [slides] [slides (in Russian)]
  • Anton Belyy, Mariia Seleznova, Aleksei Sholokhov, and Konstantin Vorontsov. Quality evaluation and improvement for hierarchical topic modeling. In 24rd International Conference on Computational Linguistics and Intellectual Technologies, pages 110–123, 2018. [paper] [slides]