Focus
Where the group works
Our aim is to develop innovative and robust natural language processing technologies that improve
machine understanding of human languages. Much of the group’s published work concentrates on
Urdu and on cross‑language English–Urdu problems, where large annotated resources have not
previously existed.
Text reuse and plagiarism detection
Detecting where text has been reused or paraphrased, both within a language and across languages.
The group has released corpora and benchmarks for Urdu news text reuse, Urdu paraphrase plagiarism
and cross‑language English–Urdu reuse, and has worked on plagiarism detection in the
MEDLINE collection.
Author identification and author profiling
Inferring characteristics of an author — such as demographics and personality traits —
from the text they write, and attributing text to authors. The group has published on author
profiling for SMS, Facebook and Twitter data, on cross‑genre profiling, and on author
diarization.
Information retrieval
Retrieving relevant documents from large collections, including query‑expansion techniques
applied to biomedical literature retrieval.
Sentiment analysis
Identifying opinion, polarity and attitude expressed in text.
Word sense disambiguation
Determining which sense of an ambiguous word is intended in a given context.
Information extraction
Extracting structured facts, entities and relations from unstructured text.
Language resources for Pakistani languages
Building the annotated corpora, lexicons and benchmarks that Pakistani languages lack, so that
standard NLP methods can be developed and evaluated for them.