Skip to content
Go to file

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time

Semantic Parsing Question Answering pipeline


This document aims to describe how to install the complete pipeline of our QA system. The pipeline currently is composed of the following 3 components

The flow of control between the three components is the following

User ----> AskNowUI ----> EARL ----> SQG
     <----          <----      <----                        

AskNowUI is the web UI component of the project which is mainly comprised of HTML and javascript code.

EARL receives the text question from the UI and after a series of steps links phrases in the question to entity and reltion URIs in a knowledge base.

SQG receives a list of entities and relations from EARL and generates optimal SPARQL queries and returns them to EARL. EARL executes these SPARQL queries to generate a final answer and returns it to the web UI.


On a computer with 8 - 10 GB of available RAM

# mkdir asknow
# cd asknow/


# git clone
# cd AskNowUI/
# git checkout debayan
# python 5001

At this point the web UI can be viewed by opening localhost:5001 on a web browser on the same machine. However for it to be useful we need to setup the next 2 components.


In a new shell

# cd asknow/
# git clone
# cd EARL/
# git checkout dev

Follow instructions at EARL github repository for installation. At the end of that process you will have EARL listening on port 4999. AskNowUI expects EARL to be at port 4999.


In a new shell

# cd asknow/
# git clone
# cd SQG/
# python

Follow the instructions on SQG github repository closely. Install any dependencies required via pip. Ensure that on running the bloom file is loaded at run time and that the dbpedia endpoint is setup correctly. For this you might have to make 2 changes. Edit kb/ __init__() function and change the variable endpoint to a locally running dbpedia sparql endpoint. Change the variable one_hop_bloom_file to ./data/blooms/spo1.bloom

Note: The version deployed on uses 30ae3eb commit sha1 of SQG.

This starts SQG at port 5000. We have completed the setup. You may now visit localhost:5001 on your web browser on the same machine and ask a question and see the three components replying.


Full Question Answering pipeline



No releases published


No packages published
You can’t perform that action at this time.