Swapna Gottipati

CQArank: jointly model topics and expertise in community question answering

Community Question Answering (CQA) websites, where people share expertise on open platforms, have become large repositories of valuable knowledge. To bring the best value out of these knowledge repositories, it is critically important for CQA services to know how to find the right experts, retrieve archived similar questions and recommend best answers to new questions. To tackle this cluster of closely related problems in a principled approach, we proposed Topic Expertise Model (TEM), a novel probabilistic generative model with GMM hybrid, to jointly model topics and expertise by integrating textual content model and link structure analysis. Based on TEM results, we proposed CQARank to measure user interests and expertise score under different topics. Leveraging the question answering history based on long-term community reviews and voting, our method could find experts with both similar topical preference and high topical expertise. Experiments carried out on Stack Overflow data, the largest CQA focused on computer programming, show that our method achieves significant improvement over existing methods on multiple metrics.

Finding relevant answers in software forums

Online software forums provide a huge amount of valuable content. Developers and users often ask questions and receive answers from such forums. The availability of a vast amount of thread discussions in forums provides ample opportunities for knowledge acquisition and summarization. For a given search query, current search engines use traditional information retrieval approach to extract webpages containing relevant keywords. However, in software forums, often there are many threads containing similar keywords where each thread could contain a lot of posts as many as 1,000 or more. Manually finding relevant answers from these long threads is a painstaking task to the users. Finding relevant answers is particularly hard in software forums as: complexities of software systems cause a huge variety of issues often expressed in similar technical jargons, and software forum users are often expert internet users who often posts answers in multiple venues creating many duplicate posts, often without satisfying answers, in the world wide web. To address this problem, this paper provides a semantic search engine framework to process software threads and recover relevant answers according to user queries. Different from standard information retrieval engine, our framework infer semantic tags of posts in the software forum threads and utilize these tags to recover relevant answer posts. In our case study, we analyze 6,068 posts from three software forums. In terms of accuracy of our inferred tags, we could achieve on average an overall precision, recall and F-measure of 67%, 71%, and 69% respectively. To empirically study the benefit of our overall framework, we also conduct a user-assisted study which shows that as compared to a standard information retrieval approach, our proposed framework could increase mean average precision from 17% to 71% in retrieving relevant answers to various queries and achieve a Normalized Discounted Cumulative Gain (nDCG) @1 score of 91.2% and nDCG@2 score of 71.6%.

Text-mining approach for verifying alignment of information systems curriculum with industry skills

Managing and developing competencies or skills are vital to professional development of an individual. Many tertiary education institutions are therefore focused on developing curriculum that will help the graduating students acquire skills that are aligned with industry practice. Industry skills frameworks such as Skills Framework for the Information Age (SFIA) define the professionals required for an IT professional. Therefore, mapping the curriculum competencies to the industry skills framework has a dual purpose of aiding the educationists to improve the curriculum, and the students to plan the courses according to their career plan. Existing mapping methods are manual and painstaking processes. In this paper, we present an automated solution based on text analytics techniques to map the curriculum to industry framework and provide a visual based analysis to discover the strengths and weakness of the curriculum. We evaluated our solution model on an undergraduate core curriculum; Bachelor of Science (Information Systems Management) degree program BSc (ISM), offered by the School of Information Systems (SIS), Singapore Management University (SMU) and Skills Framework for the Information Age (SFIA).

What's Public Feedback? Linking High Quality Feedback to Social Issues Using Social Media

In this paper we present a study of new problem of linking high quality public feedback to the issues discussed in an article. Analyzing public opinion on social and political issues as well as government policies is of particular importance to policy makers. Given a segmented article with multiple issues and public comments towards the article, the task aims to extract high quality feedback and link it to the relevant issues in the article. Our proposed solution, two-stage approach rely on supervised learning technique for extracting high quality feedback and statistical topic modeling technique for extracting the relevant feedback to the issues/topics of the article. We study the problem on two different data sets. We evaluated both the stages of the framework and the empirical results on both data sets show that the proposed approach is effective in linking high quality relevant feedback to the segments of the article.

A Retail Bank's BPM Experience

Abstract This real-life case study, which was undertaken by a leading financial services group in the Asia-Pacific region, is used to demonstrate the innovative use of BPM (Business Process Management) technology in a competitive business area. It describes how a BPM project, within the Application Verification and Capture (AVC), was conceived, designed and implemented in order to deliver strategic value to the organization. Hereafter, the financial services group will be referred to as “the bank”. The AVC project was targeted at one of the banks processes called the Application Verification and Capture (AVC) process for unit trust products. This process involved extensive paperwork and numerous manual tasks that resulted in slow processing, manual errors, rework and customer dissatisfaction. By combining process redesign and automation using information technology, the process was improved significantly.

Competency analytics tool: Analyzing curriculum using course competencies

The applications of learning outcomes and competency frameworks have brought better clarity to engineering programs in many universities. Several frameworks have been proposed to integrate outcomes and competencies into course design, delivery and assessment. However, in many cases, competencies are course-specific and their overall impact on the curriculum design is unknown. Such impact analysis is important for analysing, discovering gaps and improving the curriculum design. Unfortunately, manual analysis is a painstaking process due to large amounts of competencies across the curriculum. In this paper, we propose an automated method to analyse the competencies and discover their impact on the overall curriculum design. We provide a principled methodology for discovering the impact of courses’ competencies using Bloom’s Taxonomy, Dreyfus’ model and the learning outcomes framework. We developed the Curriculum Analytics Tool (CAT) which generates the competency scores for the entire curriculum across two dimensions; Cognitive levels and Progression levels. We use the CAT to analyse the competencies of an undergraduate Information Systems Management core curriculum program. Using 14 courses and the corresponding 578 competencies, this paper shows how our method enables us to perform in-depth analysis on the curriculum by discovering the cognition and progression statistics. We further apply the tool for recommending competencies when launching new courses.

An Integrated Model for User Attribute Discovery: A Case Study on Political Affiliation Identification

Discovering user demographic attributes from social media is a problem of considerable interest. The problem setting can be generalized to include three components — users, topics and behaviors. In recent studies on this problem, however, the behavior between users and topics are not effectively incorporated. In our work, we proposed an integrated unsupervised model which takes into consideration all the three components integral to the task. Furthermore, our model incorporates collaborative filtering with probabilistic matrix factorization to solve the data sparsity problem, a computational challenge common to all such tasks. We evaluated our method on a case study of user political affiliation identification, and compared against state-of-the-art baselines. Our model achieved an accuracy of 70.1% for user party detection task.

A Retail Bank's BPM Experience: Research Note

In a typical bank, a business process usually involves a sequence of activities that requires some combination of any of the following namely, human, hard copies of paper, soft copies of paper, hardware such as scanners and printers, and business applications. At one extreme, entirely paperbased process may involve receiving handwritten or typed information and combining it with other printed information from paper-based files or printouts from business applications. It is in this environment that process improvement initiatives must achieve goals such as reducing processing time, reducing processing effort, and improving resource utilization. A number of quantitative and qualitative process improvement techniques can be applied to optimize the process. Quantitative techniques are aimed at applying statistical methods to process execution data in order to help identify problem areas and verify improvement gains. For example, SixSigma, which was invented at Motorola in the mid 1980s, provides a set of highly sophisticated tools to analyze processes from a statistical perspective. Data is collected from processes to determine the extent of variation from target performance measurements (Subramoniam et. al., 2012). Qualitative techniques focus on analyzing the various stapes in a process flow to identify the root cause of problems and potential areas for optimization. For example, qualitative techniques such as process flow analysis and value added analysis focus on achieving performance gains by categorizing the process activities, identifying non-va1ue-adding flows or activities, and then optimizing the flows to achieve gains (Duran, 2007).

Text analytics approach to extract course improvement suggestions from students’ feedback

In academic institutions, it is normal practice that at the end of each term, students are required to complete a questionnaire that is designed to gather students’ perceptions of the instructor and their learning experience in the course. Students’ feedback includes numerical answers to Likert scale questions and textual comments to open-ended questions. Within the textual comments given by the students are embedded suggestions. A suggestion can be explicit or implicit. Any suggestion provides useful pointers on how the instructor can further enhance the student learning experience. However, it is tedious to manually go through all the qualitative comments and extract the suggestions. In this paper, we provide an automated solution for extracting the explicit suggestions from the students’ qualitative feedback comments. The implemented solution leverages existing text mining and data visualization techniques. It comprises three stages, namely data pre-processing, explicit suggestions extraction and visualization. We evaluated our solution using student feedback comments from seven undergraduate core courses taught at the School of Information Systems, Singapore Management University. We compared rule-based methods and statistical classifiers for extracting and summarizing the explicit suggestions. Based on our experiments, the decision tree (C5.0) works the best for extracting the suggestions from students’ qualitative feedback.

Deal or no deal: Catering to user preferences

A common problem in large urban cities is the huge number of retail options available. In response, a number of shopping assistance applications have been created for mobile phones. However, these applications mostly allow users to know where stores are or find promotions on specific items. What is missing is a system that factors in a users shopping preferences and automatically tells them which stores are of their interest. A key challenge in this system is building a matching algorithm that can combine user preferences with fairly unstructured deals and store information to generate a final rank ordered list. In this work, we present the first prototype of my Deal, a system that automatically ranks deals according to user preferences and presents them to the user on their mobile device.