Aparna S Varde
- Aparna Montclair CS Page
- Aparna WPI Alumni Page
- Aparna Microsoft Academic Profile
- Aparna LinkedIn Profile
- PIKM 2012: PhD workshop in ACM CIKM Conference
Decision Support in Green Information Technology
This multidisciplinary research in data mining and environmental management is supported by a grant from PSE&G. It involves investigating greener solutions for data centers with the goals of energy efficiency and adequate performance. The role played by data mining techniques is significant here in the development of a decision support system that will assist IT managers to head towards green computing in their respective data centers. This grant is supporting a Ph.D. student Michael Pawlish in Environmental Management with Dr. Varde as the dissertation advisor in her capacity as Doctoral Faculty Member in that Program. It has led to publications on in IEEE's ICIAFS, ACM's CIKM and other muldisciplinary venues.
Terminology Evolution in Information Retrieval
This research in the overall area of Web and text mining started as joint work with Max Planck Institute, Germany where Dr. Varde was a Visiting Senior Researcher. The goal of this project is to detect evolving terminology in responding to user queries on the Web by mining existing text archives. This is in order to enhance information retrieval by incorporating historical information on terms contained in queries. This led to a Masters' Project by a CS graduate student Debjani Roychoudhury and a Masters' Thesis by a CS graduate student Amal Kalurachchi. It has been published in AAAI, ACM's EDBT and ACM's CIKM conferences.
Learning By Mining Nanoscale Images
This work is funded by a grant from NSF REU ans supports undergraduate students from the tri-state area during summers. The focus of this grant is in the area of image processing and my contribution is in the area of learning from image data at the nanoscale level. The work entails proposing and implementing techniques for discovering knowledge from image data useful in domain-specific decision-making. This project involves real data obtained from researchers in Nanotechnology, used for running experiments with the proposed techniques. It has real-world applications such as drawing conclusions from biological images based on automating comparisons between them by learning suitable notions of similarity. This has the broader impact of catering to areas such as health informatics. For example, the results of the learning process are useful in finding a cheaper material instead of a more expensive material to develop a human body implant, if both materials yield similar results as evident from image similarity search. This is given the fact that these images are generated from real experimental Publications from this work include a paper in SPIE 2010 conference, a presentation in ACM CCSC 2010 conference, a paper in ICML 2010 Workshops and two earlier papers in IEEE ICDE DBRank workshop and the ICDCIT conference.
Treatment of Article and Collocation Errors in L2 English Texts
This research in the general area of text mining and computational linguistics involves the detection, correction and causal analysis of errors in article usage and collocation that occur in texts written by L2 (non-native) English speakers. Article errors pertaining to entering articles where not needed, omitting articles where needed, entering the wrong article. Collocation errors involve using incorrect collocation such as powerful tea when the user actually means strong tea. We are proposing methods to detect such errors, analyze their causes and suggest correct responses to them. Mining the concerned text and deploying machine learning techniques such as classification and ensemble learning play an important role here. Related publications include a paper accepted in the AAAI based FLAIRS 2010 conference on classification of article errors and one more in submission to a suitable conference on collocation error correction. A grant is in submission in for this work.
XML-based Markup Langauges and Cloud Computing in EHR Management
This work is supported by a grant to fund Honors students in BS degree programs in various science disciplines who are expected to work with their respective faculty mentors to complete an undergraduate thesis in a concerned area. My role as faculty mentor is to work with a BS student in Information Technology on a specific research project, namely, XML-based markup languages and Cloud Computing in management of EHR (Electronic Health Records). We are investigating the use of the medical markup language MML for storing and exchanging health records, proposing techniques for knowledge discovery over such XML based standards and also investigating the use of cloud technology in storage, retrieval and knowledge discovery pertaining to healthcare taking into account issues such as cost-effectiveness, risk analysis and scalability. Related publications include a paper in the IEEE ICDM 2011 conference in their KDCloud workshop and one more in submission to a suitable journal.
Scientific Data Management and Mining with Cloud Computing
This project focuses on research in cloud computing with emphasis on managing and mining large scale scientific data. Besides a thorough investigation of existing methodologies, it addresses the design and implementation of novel techniques and the enhancement of existing approaches for scientific data analysis on the cloud. Risk assessment forms an important aspect of this work. The grant proposal also entails the development of objective metrics to evaluate the performance of cloud technologies with particular emphasis on scientific data, considering issues such as efficiency, accuracy, complexity and scalability. It also addresses the prospects of greening the environment through the use of cloud computing, considering the contributions that can be made by the cloud with respect to greenness and evaluating the same subjectively and objectively, with a broader impact on sustainability. This has led to a Masters' Project by the student Shireesha Chandra and is in submission to suitable venues. More work is in progress with the potential involvement of other students.