Hello! I'm Ayushi Dalmia

Graduate Student, Search and Information Extraction Lab, IIIT Hyderabad

Striving to connect the dots

Recently I have joined IBM Research, Delhi. Prior to this I was a graduate student at Search and Information Extraction Lab, IIIT, Hyderabad. My advisors are Prof. Manish Gupta and Prof. Vasudeva Varma. I am yet to defend my thesis.

My research interests are in the areas of Data Mining and Machine Learning in general and its application to Complex Network Analysis in particular. More specifically, I focus on outlier detection and community detection in complex networks. I am also enthusiastic about Natural Language Processing.

I did my Bachelor's in Computer Science and Engineering from Heritage Institute of Technology, Kolkata. During my undergraduation, I did a summer internship at CNeRG, IIT Kharagpur where I worked on citation networks with Dr. Niloy Ganguly . Dr. Tanmoy Chakraborty was my mentor there. As part of my final year project, I worked on recommender systems under the supervision of Dr. Prosenjit Gupta and Dr. Subhasish Majumder.

Prior to this I completed high school from St. John's Diocesan Girls' Higher Secondary School

Education

Publication Summary

Select Projects

  • Permanence for Local Community Detection

    Working with Dr. David Gleich and Dr. Tanmoy Chakraborty to formulate a metric for local community detection using the principle of permanence, an established metric for non-overlapping community detection.
  • Metrics for Community Analysis: A Survey

    Working with Dr. Tanmoy Chakraborty, Dr. Niloy Ganguly and Dr. Animesh Mukherjee to submit a survey paper to summarise, the state-of-the-art metrics proposed for the detection and the evaluation of community structure.
  • Research Challenges in Bibliographic Dataset: A Linguistic Perspective

    Working with Dr. Tanmoy Chakraborty on a challenge paper to summarise the research challenges involved for various natural language processing task in bibliographic text as the performance of the state-of-the-art approaches in resolving these problems is inadequate owing to the challenges presented by the nature of scientific articles.(under review at ACM Journal of Data and Information Quality)
  • Sandhan, Cross Language Information Access

    Participated in the RecSys 2014 challenge where the task is to predict the user engagement in tweets (retweet and favorite) for items.The approach involved extraction of interesting hand-crafted features, including usage of external data from IMDb and experimenting with various supervised learning techniques.
  • Graph Analysis and Community Detection in Citation Network

    Worked on a novel algorithm OverCite, for Overlapping Community Detection in a Citation Network where it is viewed as a tripartite hypergraph. It was later published as a paper at ASONAM 2013. Generated a network after removing self citations and collaborative citations from a citation network of 1 million papers followed by comparison of the network properties for the original and the generated network.
  • Survey on Event Detection

    An extensive survey on Event Detection techniques in Social Media. More specifically, this survey covered the techniques used to detect events in Twitter. Report
  • Drop me an Email

    ayushi [dot] dalmia [at] research [dot] iiit [dot] ac [dot] in