Name: Rishiraj Saha Roy
Type: User
Company: Fraunhofer IIS
Bio: Rishiraj Saha Roy is an NLP Researcher at the Fraunhofer IIS, Erlangen, Germany. He works on applications of LLMs.
Location: Saarbruecken, Germany
Blog: https://www.linkedin.com/in/rishiraj-saha-roy-35736182/
Rishiraj Saha Roy's Projects
CLOCQ is a framework which allows efficient access to knowledge bases (KB) for functionalities related to question answering (QA). CLOCQ can retrieve a set of relevant facts from the KB for a given user question. Further, it provides efficient retrieval of KB-neighborhoods, KB-connectivities and access to metadata such as labels, aliases or types.
Code for our SIGIR'21 paper "Reinforcement Learning from Reformulations in Conversational Question Answering over Knowledge Graphs"
CONVEX is the first completely unsupervised method to deal with Conversational Question Answering over Knowledge Graphs.
Code for our submission to the Conversational Assistance Track of the Text Retrieval Conference (TREC CAsT) 2019
ELIXIR: Learning from User Feedback on Explanations to Improve Recommender Models
Code for our CIKM'21 paper "Complex Temporal Qestion Answering on Knowledge Graphs"
Code for our SIGIR 2023 paper. EXPLAIGNN provides a pipeline for conversational question answering (ConvQA) over heterogeneous sources, and code for iterative graph neural networks (GNNs). Such iterative GNNs can help to causally explaignn GNN outputs.
Automatically exported from code.google.com/p/graphlabapi
Automatically exported from code.google.com/p/graphlabapi.graphlab3
This repository contains all the code relevant to the query language model syntactic complexity project.
Code for our WSDM'24 paper "Robust Training for Conversational Question Answering Models with Reinforced Reformulation Generation"
Incomplete follow-up questions dataset used for comparison with SIRI for SIGIR 2017 paper
Tools for the TREC CAsT benchmark
Project to prepare a n-triples wikidata dump for QA access.