sebastian-hofstaetter / neural-ir-explorer

Neural-IR-Explorer: A Content-Focused Tool to Explore Neural Re-Ranking Results
https://neural-ir-explorer.ec.tuwien.ac.at/
Apache License 2.0
33 stars 3 forks source link
interactive interpretability neural-ir ranking transformer

Neural-IR-Explorer

Sebastian Hofstätter, Markus Zlabinger and Allan Hanbury ECIR 2020

https://arxiv.org/abs/1912.04713

The content-focused Neural-IR-Explorer empowers users to browse through neural re-ranking results and inspect the inner workings and fine-grained results of neural re-ranking models. We aim to explore the reasons behind ranking results.

The explorer includes a categorized overview of the available queries, as well as an individual query result view with various options to highlight semantic connections between query-document pairs.

Features

Query overview:

Single query result view:

Diff of two documents for a single query:

Run the web-server

set RUN_CONFIG=config_knrm.yaml
set FLASK_APP=api.py
flask run

Run the webpack builder

npm run build

Dependency on neural re-ranking results

TODO - We will add a reference to our neural library as soon as it is public