Research & publications
My research started with a simple question: how do you get a machine to read many documents and answer in a few good sentences?
Question answering, multi-document summarisation, retrieval and entailment-based automatic evaluation — the work of my Ph.D. years — tackle the same problems as today’s retrieval-augmented generation, LLM evaluation and semantic search. At Samsung the same questions moved onto phones and into images.
- publications
- 35+
- citations
- 460+
- h‑index
- 11
Source: Google Scholar, September 2026
Selected publications
-
POP-VQA – Privacy Preserving, On-Device, Personalized Visual Question Answering
WACV 2024
On-device VQA personalised with a user knowledge graph; +36% absolute on KVQA 1‑hop questions.
Topics: Multimodal & on-device Knowledge graphs
-
Multi-Hop Relation Aware Representations for Inductive Knowledge Graphs
ICON 2023 pp. 27–35
Topics: Knowledge graphs
-
From Literature to Knowledge: Exploiting PubMed to Answer Biomedical Questions in Natural Language
ITBAM 2015 · Springer LNCS pp. 3–15
Biomedical question answering over PubMed; from the ERCIM postdoctoral fellowship at IIT‑CNR, Pisa.
Topics: Question answering
-
Keyphrase Extraction in Scientific Articles: A Supervised Approach
COLING 2012 · Demonstration Papers pp. 17–24
Topics: Language tools
-
Question Answering System for QA4MRE@CLEF 2012
QA4MRE @ CLEF 2012 Working Notes
Top-ranked system, c@1 = 0.65.
Topics: Question answering
-
A Query Focused Multi Document Automatic Summarization
PACLIC 24 (2010) pp. 545–554
Best-paper nominee, student session.
Topics: Summarization
Shared-task results
| Evaluation campaign | Task | Result |
|---|---|---|
| CLEF 2013 | QA4MRE — Question Answering for Machine Reading Evaluation | Top-ranked system · c@1 0.59 |
| CLEF 2012 | QA4MRE | Top-ranked system · c@1 0.65 |
| CLEF 2011 | QA4MRE | Top-ranked system · c@1 0.57 |
| CLEF 2010 | ResPubliQA — answer selection | Top-ranked · 0.63 |
| INEX 2011 | Question Answering track | Top on the relaxed readability metric |
Also participated: INEX Tweet Contextualization 2012–13 · FIRE ad-hoc retrieval 2008/2010/2012 · HOO 2011–12 (grammatical error correction) · TAC RTE-7 2011 · PAN 2011 (plagiarism detection) · TAC summarisation 2009–10 · SemEval-2010 keyphrase extraction.
All publications
37 items, newest first · filter by type, topic or keyword
-
POP-VQA – Privacy Preserving, On-Device, Personalized Visual Question Answering
WACV 2024
On-device VQA personalised with a user knowledge graph; +36% absolute on KVQA 1‑hop questions.
Topics: Multimodal & on-device Knowledge graphs
-
Multi-Hop Relation Aware Representations for Inductive Knowledge Graphs
ICON 2023 pp. 27–35
Topics: Knowledge graphs
-
An On-device Deep-Learning Approach for Attribute Extraction from Heterogeneous Unstructured Text
ICON 2021 pp. 573–582
Topics: Multimodal & on-device Knowledge graphs
-
From Literature to Knowledge: Exploiting PubMed to Answer Biomedical Questions in Natural Language
ITBAM 2015 · Springer LNCS pp. 3–15
Biomedical question answering over PubMed; from the ERCIM postdoctoral fellowship at IIT‑CNR, Pisa.
Topics: Question answering
-
Cross Lingual Snippet Generation Using Snippet Translation System
CICLing 2014 · Springer LNCS 8404 pp. 331–342
Topics: Retrieval & search
-
A Hybrid Question Answering System for Multiple Choice Question (MCQ)
QA4MRE @ CLEF 2013 Working Notes
Top-ranked system, c@1 = 0.59.
Topics: Question answering
-
Answering Questions from Multiple Documents – the Role of Multi-Document Summarization
RANLP 2013 Student Research Workshop pp. 14–21
Topics: Question answering Summarization
-
Automatic Evaluation of Summary Using Textual Entailment
RANLP 2013 Student Research Workshop pp. 30–37
Topics: Evaluation & entailment Summarization
-
Entailment-based Fully Automatic Technique for Evaluation of Summaries
Research in Computing Science, vol. 65 pp. 11–23
Topics: Evaluation & entailment Summarization
-
Multi-Document Summarization using Automatic Key-Phrase Extraction
RANLP 2013 Student Research Workshop pp. 22–29
Topics: Summarization
-
Multiple Choice Question (MCQ) Answering System for Entrance Examination
QA4MRE @ CLEF 2013 Working Notes
Topics: Question answering
-
Query Focused Language Independent Multi-document Summarization and Information Retrieval for English and Bengali
LAP LAMBERT Academic Publishing
M.Tech. thesis published as a monograph.
Topics: Summarization Retrieval & search
-
Transliterated Search System for Indian Languages
FIRE 2013 Working Notes
Topics: Retrieval & search
-
Tweet Contextualization (Answering Tweet Question) – the Role of Multi-document Summarization
INEX @ CLEF 2013 Working Notes
Topics: Summarization
-
A Hybrid QA System with Focused IR and Automatic Summarization for INEX 2011
INEX 2011 · Springer LNCS 7424 pp. 207–218
Topics: Question answering Summarization
-
Comparative & Evaluative QA System in Tourism Domain
CCSEIT 2012 · ACM ICPS pp. 458–465
Topics: Question answering
-
Keyphrase Extraction in Scientific Articles: A Supervised Approach
COLING 2012 · Demonstration Papers pp. 17–24
Topics: Language tools
-
Language Independent Query Focused Snippet Generation
CLEF 2012 · Springer LNCS 7488 pp. 138–140
Topics: Retrieval & search Summarization
-
A Hybrid Tweet Contextualization System using IR and Summarization
INEX @ CLEF 2012 Working Notes
Topics: Summarization Retrieval & search
-
Ad-hoc Information Retrieval focused on Wikipedia based Query Expansion and Entropy Based Ranking
FIRE 2012 Working Notes
Topics: Retrieval & search
-
An Automatic System for Modality and Negation Detection
QA4MRE @ CLEF 2012 Working Notes
Topics: Question answering
-
Answer Extraction of Comparative and Evaluative Question in Tourism Domain
IJCSIT, vol. 3(4) pp. 4610–4616
Topics: Question answering
-
Cross Lingual Query Dependent Snippet Generation
IJCSIT, vol. 3(4) pp. 4603–4609
Topics: Retrieval & search
-
Detection and Correction of Preposition and Determiner Errors in English: HOO 2012
BEA-7 @ NAACL-HLT 2012 pp. 201–207
Topics: Language tools
-
Question Answering System for QA4MRE@CLEF 2012
QA4MRE @ CLEF 2012 Working Notes
Top-ranked system, c@1 = 0.65.
Topics: Question answering
-
A Hybrid Question Answering System based on Information Retrieval and Answer Validation
QA4MRE @ CLEF 2011 Notebook Papers
Top-ranked system, c@1 = 0.57.
Topics: Question answering
-
A Rule Based Approach for Analysis of Comparative or Evaluative Questions in Tourism Domain
KRAQ @ IJCNLP 2011 pp. 29–37
Topics: Question answering
-
A Textual Entailment System using Anaphora Resolution
TAC 2011 · RTE-7 (NIST)
Topics: Evaluation & entailment
-
May I check the English of your paper!!!
HOO @ ENLG 2011 pp. 250–253
Topics: Language tools
-
Rule Based Plagiarism Detection using Information Retrieval
PAN @ CLEF 2011 Notebook Papers
Topics: Language tools
-
A Query Focused Automatic Multi Document Summarizer
ICON 2010 pp. 241–250
Topics: Summarization
-
A Query Focused Multi Document Automatic Summarization
PACLIC 24 (2010) pp. 545–554
Best-paper nominee, student session.
Topics: Summarization
-
Bengali Parsing System at ICON NLP Tool Contest 2010
ICON 2010 NLP Tools Contest pp. 20–24
Topics: Language tools
-
JU_CSE_TE: System Description QA@CLEF 2010 - ResPubliQA
ResPubliQA @ CLEF 2010 Notebook Papers
Top-ranked in answer selection, score 0.63.
Topics: Question answering Evaluation & entailment
-
Theme Based English and Bengali Ad-hoc Monolingual Information Retrieval in FIRE 2010
FIRE 2010 Working Notes
Topics: Retrieval & search
-
Dependency Parser for Bengali: the JU System at ICON 2009
ICON 2009 NLP Tools Contest pp. 7–11
Topics: Language tools
-
English Bengali Ad-hoc Monolingual Information Retrieval Task Result at FIRE 2008
FIRE 2008 Working Notes
Topics: Retrieval & search
No publications match those filters.