My research investigates information seeking behaviors of individuals and groups in various collaborative and social contexts, including education, libraries, and work environments. Collaborative information seeking (CIS) involves a coordinated effort to better understand and solve a shared problem or interest, such as siblings looking for information about an elderly parent’s medical condition. In contrast, Social Information Seeking (SIS) refers to queries or questions people pose to their peers or a community to obtain their thoughts, opinions, and validations. One example of this would be asking for travel advice on an online community-based question-answering service such as Yahoo! Answers. My scholarship involves the development of research models and methods that extend the traditional single-user view of information seeking in SIS and CIS contexts. My research in these areas is informed by, and contributes to basic research of Interactive Information Retrieval (IIR) at the level of the individual interacting with information, and to developments and innovations in Social Media/Networking related to information access and usage at society/community level.
Below I describe the four current themes of my research, my approaches, and the findings so far. More description about various research projects can be found here. I have published and talked about these works extensively. All the tools developed through my research work are provided for free under open-source model.
Collaborative Information Seeking (CIS)
My research on collaborative information seeking (CIS) has been primarily concerned with investigating how and why people work in collaboration in information-intensive tasks, and what support can be provided to facilitate and encourage such collaborative behavior. This topic has been my primary focus of research dating to my PhD dissertation , subsequently developed into a book, and most recently a special issue of IEEE Computer Magazine . On a theoretical side, through these works, I have proposed, developed, tested, and rectified C5 Model of Collaboration.
My research work includes interview-based, laboratory, and field studies to investigate various aspects of CIS behaviors and applications. My interviews and design sessions have provided insights into motivations and difficulties people encounter when they collaborate. My laboratory studies involving hundreds of participants (individuals or teams) have allowed us to extend our knowledge about how communication and awareness affect CIS , and how we could design systems to create synergic effect (whole being greater than the sum of the parts) based on that knowledge. For instance, using a new algorithm that intelligently mediates collaborative actions among the participants of a collaborative project, I have shown how we could achieve both relevance and novelty/diversity in information retrieval.
I have developed Coagmento, a system for studying individual and collaborative information seeking in lab and field settings. Using this system and several experiments, I have been able to extend our knowledge about CIS in various time and space contexts in exploratory search tasks. For instance, concerning the space dimension, using controlled lab experiments I showed that remotely located collaborators have more diversity and better discovery of information, whereas co-located collaborators tend to have more social interactions. Regarding the time dimension, I found that synchronously working collaborators do division-of-labor around the topic, whereas asynchronously working collaborators apply the strategy of division around the process such as searching and writing.
Social Information Seeking (SIS)
My research on social information seeking (SIS) has relied primarily on studying community-based question-answering (Q&A) services such as Yahoo! Answers and WikiAnswers.
I have built Web-crawlers to collect massive amounts of data from such services, resulting in a collection of tools, data, and methods for content analysis. This has resulted in complete dataset for Google Answers, and datasets for Yahoo! Answers and Stack Overflow, both containing millions of questions, answers, and user profiles. Through analyses of such content, I have been able to show how and why an information sharing community succeeds or fails. In addition to content analysis, I have also conducted several surveys and interviews in an attempt to learn about how and why people seek information from such social/community-based forums. We have learned that contextual factors associated with the answerers, such as the number and the quality of past answers and questions they posted, are often more indicative than the content of an answer in regard to how well that answer will be received.
On the theoretical side, I have developed a new typology of question types within different Q&A services. These are now being used to understand why a question fails to receive a good or any answer and what can be done to address this failure. I have also compiled a comprehensive list of aspects, subdivided into relevance, quality, and satisfaction, that contribute to the goodness of an answer. This list has been utilized for understanding the difference in answering behaviors between an expert-driven virtual reference services and user-driven social Q&A services, as well as for building statistical models that could evaluate the quality of an answer in order to make suggestions to improve it.
Interactive Information Retrieval (IIR)
To better understand people’s information seeking processes and to provide personalization and recommendations to them as methods for helping the information seekers, I have been involved in several research activities that investigate elements of interactive information retrieval (IIR), specifically focusing on people’s intentions. Using data from eye-tracking behaviors (eye-gazing, eye-fixation) as well as general Web behaviors (query formulation, page visitations), we identified 6-8 behavioral clusters within IIR. Different patterns of such behaviors indicate different motivating search tasks.
When it comes to personalization and recommendations in an IIR situation, I have taken an approach that is not often common in the literature; instead of simply looking at the outcomes (queries, documents, snippets) of an IIR episode, I have been trying to understand the underlying process that a user goes through, use that information to identify potential problems in the future, and provide strategic recommendations to correct the search path before it is too late. I have developed a new method for doing such process-based evaluation and recommendation in IIR, which was tested using data collected through four different user studies involving 216 users with strong positive results. The findings, and a new framework that resulted from this approach and experiments are reported in a new JASIST article .
I have also been looking at various aspects of social media to address research questions relating information seeking and sharing through social/community channels such as YouTube, Twitter, and Facebook. Much of my work in this area has been interdisciplinary in nature, in that I have been collaborating with researchers from various fields such as political science, media studies, journalism, archiving and preservation, and information and computer sciences
To facilitate my research on social media and related topics, I have developed tools such as ContextMiner, InfoExtractor, TubeKit, and SOCRATES. These tools have been extensively used not only by me, but also a broader research community around the world for collecting large amounts of data from different social media venues for studying various social, economic, and political issues. In 2010, ContextMiner won the best political science software award from the American Political Science Association. This accolade recognized ContextMiner’s role in helping political scientists and journalists studying various socio-political issues through the use of social media. An article describing such a study that I co-authored received the best paper award from the International Communication Association (ICA) in 2011.
My most recent work with social media/networking is associated with SOCRATES, which is allowing me to not only build more advanced tools for collecting, analyzing, and exploring social media data, but also start creating communities of researchers and practitioners around these issues.