Ashutosh Jadhav

From Knoesis wiki
Revision as of 20:44, 30 July 2015 by Ashutosh (Talk | contribs) (Technical Skills)

Jump to: navigation, search
Ashu-2.png

Graduate Research Assistant - Kno.e.sis - Ohio Center of Excellence in Knowledge-enabled Computing
Ph.D. Candidate - Computer Science and Engineering Wright State University
Advisor - Dr. Amit Sheth


LinkedIn profile
Google Scholar
Bibliography on NCBI PubMed


Research interest: Search Intent Mining, Social Media Analytics, Health Informatics, Text Analytics, Semantic Web, Search Log Analysis, Knowledge Extraction and Utilization


Education and Work Experience

Education

  • Ph.D. Candidate, Computer Science and Engineering, Wright State University, April 2009 - Present
    • Big & Smart Data Certification from Computer Science Department, Wright State University, Dayton, OH


Work Experience

  • Research Intern, Health Science Research, Mayo Clinic, Rochester, MN (May 2013 - May 2014)
    • 1) Developed a consumer-centric health categorization framework using Semantic Web techniques 2) Semantic Intent mining for health search, based on rich knowledge from UMLS and novel approach for semantic similarity computation 3) Qualitative analysis of online health information seeking using data science techniques
    • Mentor Dr.Jyotishman Pathak
  • Software Developer Intern, Speedway SuperAmerica LLC, Enon, Ohio (June 2008 – Sept 2008)
    • Worked with credit card transaction department on a database transaction integrity project involving the implementation of scripts to automatic verification and integrity/quality checks.


Research Projects

Search Intent Mining

Search Intent Mining Overview
  • Motivation
    • Understanding intent from search queries is crucial for designing an intelligent search engine. The objective of this project is to identify health information seeking intent from queries.
  • Approach
    • First, we conducted three qualitative focus group studies to get users’ perspective on online health information seeking. Second, we selected 14 consumer-oriented health search intent classes based on inputs from focus group studies and based on analysis of popular health websites, literature survey, and empirical study.
    • Finally, I developed a semantic-driven rule-based intent framework for the identification of health seeking intent in disease agnostic manner. The framework is based on rich background knowledge from UMLS and utilizes UMLS semantic types and concepts.
    • I developed the framework using Big Data technologies such as Hadoop-MapReduce, Hive, and HBase to process millions of search queries efficiently.
  • Real World Application
    • This framework will be integrated at Mayo Clinic to create user interest profiles based on their search history. The user profiling will be further used for personalized health information intervention, content recommendation, and targeted advertisements

Semantic Query Expansion

SemanticQueryExpansion.png
  • Motivation
    • The first step in search query processing is to understand user’s information need from the submitted search query and reformulating a seed query to improve retrieval performance (Query Expansion).
    • Multiple IR studies have shown that a search engine’s performance degrades with an increase in the complexity of the search query.
  • Approach
    • In this project, I am specifically focusing on semantic expansion of complex health search queries (queries with multiple health concepts).
    • I have used hyponymy and hypernymy relationships between health concepts to expand complex health search queries.
    • I am extending current query expansion approaches by leveraging:
      • Health domain semantics,
      • Consumer health vocabulary,
      • UMLS concept hierarchy, and
      • Semantic similarity between health concepts (hyponymy and hypernymy).
    • Examples

Social Health Signals

Social Health Signals
  • Objective
    • The objective of this project is to understand and satisfy users’ need for keeping track of new information in the healthcare and well-being domain.
    • The project harvests collective intelligence to aggregate high quality, reliable, and informative healthcare content shared over social media on one platform, Social Health Signals (SHS).
  • Approach
    • We utilized a hybrid approach based on rule-based filtering and supervised Machine Learning classification to facilitate identifying informative health-related information.
  • Tool capabilities
    • Retrieve relevant and reliable health information shared on Twitter in real-time.
    • Question answering on Twitter data.
    • Rank results based on relevancy, popularity and reliability.
    • Efficient browsing of the results, we semantically categorize the information into health categories such as symptom, food and diet, healthy living, and prevention.

Twitris- Social Media Analytics and Research Platform, Demo

Twitris in action for depression use-case
  • Objective
    • Twitris addresses information overload problem in social media by analyzing of large-scale social data and facilitates understanding of the social media content.
  • Tool capabilities
    • Twitris uses state-of-the-art technologies to
      • Process real-time large scale social media streams,
      • Extract meaningful signals from noisy social media data,
      • Semantically integrate multiple complimentary web resources such as images, news, videos, and Wikipedia articles,
      • Provide actionable insights, and
      • Visualize processed information in Spatio-Temporal-Thematic, People-Content-Network and Sentiment-Emotion-Intent dimensions.
  • My Role
    • I coordinated the development, analysis, and UI design for Twitris and Twitris 2.0. I contributed to the text analysis, information extraction, and information integration components of the system.
  • Real World Application
    • Twitris system has been used in multiple real world scenarios such as brand tracking (e.g. Apple, Macy’s), review monitoring (e.g. iPhone, movies), public event monitoring (e.g. US presidential election, state fairs), disaster situations (e.g. Japan earthquake) and health monitoring (e.g. Swine flu, depression).



Device Effect on Online Information Seeking

  • Motivation
    • Personal computers (desktops, laptops) and smart devices (smartphones, tablets) have distinct characteristics in terms of readability, user experience, accessibility, etc.
    • As search traffic from smart devices is exponentially increasing, it is critical to understand the effects of the device used for online information seeking.
    • Such knowledge can be applied to improve the search experience and to develop more advanced next-generation knowledge and content delivery systems.
  • Approach
    • In this study, I used data science techniques to perform comparative analysis (such as structural, linguistic, textual and semantics analysis) of large-scale (more than 100 millions) health search queries submitted through Web search engines from both types of devices.


Context-aware Content Recommendation in Enterprise Social Network

Research Internship at HP Services Research Lab, HP Labs, CA
Claudio Bartolini (manager) and Hamid Motahari (mentor)

  • Objective
    • During my HP Labs internship, I worked on the ‘Semantic RFF Recommendation’ to expedite the Request for Proposal (RFP) response process at HP by making the right recommendations (past RFP and deal documents) at the right time.
  • Approach
    • I developed a recommender system using a semantic document similarity approach leveraging information about content (current and past documents), enterprise social networks (people and their roles), and stages of the RFP.

Based on my internship work, HP Labs filed a patent (83138852) on Context-Aware Deal Recommendations with myself being primary inventor.


Disaster Situation Awareness based on Social Media Analysis

  • Objective
    • The objective of this project was to expedite the decision-making process in disaster situations by identifying location-based, actionable information from social media.
  • My Role
    • I lead the research, design, and development for the project (funded by the Air Force Research Lab (AFRL)) to create a geo-social mash-up for situational awareness in a disaster situation based on social media data analysis.


Wikipedia Knowledge Extraction

  • Objective
    • This project aimed to generate a topic-specific domain model based on information available from Wikipedia.
  • My Role
    • I worked on the extraction of web-links (Wikipedia links and web URLs) mentioned on the whole corpus of Wikipedia articles as a part of Doozer, a tool for automatically creating a topic-specific domain model using curated knowledge from Wikipedia.



Technical Skills

  • Programming: Java, PHP, HTML, XML, Shell scripting
  • Big Data Technologies: Hadoop, MapReduce, Apache Storm, Spark
  • Databases: MySQL, MS Access, Virtuoso triple store
  • Semantic Web: OWL, RDF, SPARQL, Ontology
  • Research Tools: Weka, Protege, Lucene, UMLS MetaMap, Google Analytics, IBM NetInsight
  • Operating Systems: Mac OS, Linux, MS Windows


Big & Smart Data Certification from Computer Science Department, Wright State University, Dayton, OH

  • To manage and analyze Big Data and to create Smart Data enabled applications for enterprises and individuals.


Publications

Google Scholar citation count: 128

Bibliography on NCBI PubMed


Journal Papers and Book Chapters


Conference Papers

  • Social Health Signals
    • Ashutosh Jadhav, Swapnil Soni, Amit Sheth
    • International Semantic Web Conference (ISWC 2015), Pennsylvania, USA, October 11-15, 2015 (submitted)
  • Twitris: Socially Influenced Browsing
    • Ashutosh Jadhav, Wenbo Wang, Raghava Mutharaju, Pramod Anantharam, Vinh Nguyen, Amit P. Sheth, Karthik Gomadam, Meenakshi Nagarajan, and Ajith Ranabahu
    • Semantic Web Challenge 2009, 8th International Semantic Web Conference, Oct. 25-29 2009, Washington, DC, USA


Posters


US Patent

  • Context-Aware Information Recommendation, filed in January 2013 by HP Labs
    • Patent filled based on HP summer 2011 internship work
    • Ashutosh Jadhav, Hamid Motahari, Susan Spence, Claudio Bartolini (Role: Primary Investigator)


Research Grants and Proposals

(Significant contribution)

  • NIH-R01 proposal (recommended for funding by review committee)
    • Modeling Social Behavior for Healthcare Utilization and Outcomes in Depression
    • In collaboration with Mayo Clinic and partly based on my Mayo Clinic internship work


  • Air Force Research Lab (AFRL) proposals
    • SIDFOT (Sensors Integration for Data Fusion in Operations and Training) project
      • Title: Geo-Social mash-up for situational awareness in a disaster response situation
      • Funded project: 2010-2011, Real-time Twitris
    • Information Operations/Cyber Exploitation Research (ICER) Program, City Beat
      • Title: Social media analysis for situational awareness
      • Funded project: 2011-2012
    • WBI's Tec^Edge Innovation and Collaboration Center (Tec^Edge ICC)
      • Funded project: Summer 2010, Summer 2011


  • Mayo Clinic Meritorious Awards
    • Healthcare trend surveillance using social networks and health search queries (funded 2013)
    • What makes a health-related tweet informative – patients’ perspective (funded 2014)


Selected Coursework

  • Semantic Web
  • Cloud Computing
  • Data Mining
  • Web 3.0 and Social Semantic Web
  • Machine Learning
  • Knowledge Representation for the Semantic Web
  • Parallel Programming with MPI
  • Distributed Computing Principles
  • Web Information System
  • Data Structures and Algorithms
  • Database Systems and Design
  • Advance Database Systems
  • Computer Engineering Mathematics
  • Computer Vision
  • Advanced Computer Networks
  • Multimedia Coding and Communication
  • Comparative languages
  • Computer Architecture I and II


Mentorship

  • Nishita Jaykumar
    • Computer Science Masters
    • Project in Social Health Signals
    • Jan. 2013 – Apr. 2013
  • Sreeram Vallabhaneni
    • PhD student
    • Project: Semantic understanding of health related post shared on twitter
    • Apr. 2012 – Aug. 2012
  • Michael Cooney
    • CS Undergrad
    • Projects: Twitris, Disaster Situation Awareness
    • 2010 – 2011


Professional Activities

  • Reviewer AMIA Annual Symposium 2015
  • Subreviewer for ICWSM 2011, 2012, 2013, 2014, 2015
  • External sub-reviewer for World Wide Web 2013, ‘Social Networks and Graph Analysis’ track
  • External reviewer for IEEE Internet Computing Magazine
  • External reviewer for Semantic Web Journal


Selected Presentations and Talks

  • Consumer-centric Health Informatics and Social Media Analytics, Mayo Clinic, Rochester, MN, USA, May 2015
  • Semantic Analysis of Online Health Information Seeking for Cardiovascular Diseases, AMIA Annual Symposium, Washington DC, USA 2014
  • Semantic Intent Mining for Health Search, European Medical Informatics Conference (MIE 2014), Istanbul, Turkey
  • Context-Aware Content Recommendation In Enterprise Pursuit Process, HP Labs, Palo Alto, CA, USA 2011
  • Enterprise Social Network: State of the Art, presented at HP Labs, Palo Alto, CA, USA 2011

Contributed significantly in the following presentations


Contact Information

  • Ashutosh Jadhav
    • Email: ashutosh@knoesis.org
    • 380 Semantic Web lab, Joshi Research Center, Wright State University, 3640 Colonel Glenn Hwy, Dayton, Ohio 45435-0001
    • LinkedIn


Ashutosh Jadhav Resume