Our group meets on Thursdays at 11:00 a.m., FL 246.
Dates and topics for this semester are as follows:
- Sept 9
- Organizational Meeting (Coordinator: Prof. Carolina Ruiz)
- Sept 16
- Joining the CS Theory Seminar:
Prof. Dan Dougherty
Graphs and the Semantic Web
In this talk I will discuss some recent results of Claudio Gutierrez, Carlos Hurtado, and Alberto Mendelzon, as presented in their paper "Foundations of Semantic Web Databases" by (Principles of Database Systems 2004).
- Sept 23
- Yoonsoo Kim
MS Thesis Presentation: Addressing the Data Recency Problem in Collaborative Filtering Systems
Recommender systems are being widely applied in many E-commerce sites to suggest products, services, and information items to potential users. Collaborative filtering systems, the most successful recommender system technology to date, help people make choices based on the opinions of other people. While collaborative filtering systems have been a substantial success, there are several problems that researchers and commercial applications have identified: the early rater problem, the sparsity problem, and the large scale problem. Moreover, existing collaborative filtering systems do not consider data recency. For this reason, if a user's preferences have changed over time, the systems might not recognize it quickly. This thesis studies how to apply data recency to collaborative filtering systems to get more predictive accuracy. We define the data recency problem as the negative impact of old data on the predictive accuracy of collaborative filtering systems. In order to mitigate this shortcoming, the combinations of time-based forgetting mechanisms, pruning and non-pruning strategies and linear and kernel functions, are utilized to apply weights. A clustering technique is employed to detect the user's changing preferences. We apply our research approach to the DeliBook dataset. The goal of our experiments is to show that our algorithm that incorporates temporal factors provides better recommendations than existing methods.
- Sept 30
- Prof. Carolina Ruiz
Research Projects in Knowledge Discovery and Data MiningIn this talk I will provide a brief tutorial-like introduction to the fields of knowledge discovery in databases and data mining as well as a description of some of the ongoing research projects at the WPI Knowledge Discovery in Databases and Data Mining Research Group (KDDRG).
- Oct 7
- Prof. Neil Heffernan
Research in Intelligent Tutoring Systems
- Oct 14
- Janet Elizabeth Burge
An Integrated Approach for Software Design Checking Using RationaleDesign Rationale (DR), the reasons behind decisions made while designing, offers a richer view of both the product and the decision-making process by providing the designer's intent behind the decisions. DR is also valuable for checking to ensure that the intent was adhered to throughout the design, as well as pointing out any unresolved (or undocumented) issues that remain open. While there is little doubt of the value of DR, it is typically not captured during design. SEURAT (Software Engineering Using RATionale) is a system we have developed to explore uses of design rationale. It supports both the display of and inferencing over the rationale to point out any unresolved issues or inconsistencies. SEURAT is tightly integrated with a software development environment so that rationale capture and use can become integrated into the software development process.
- Oct 21
- No meeting this day - Term break
- Oct 28
- Goss Nuzzo-Jones
Applying Machine Learning Techniques to Rule Generation in Intelligent Tutoring SystemsThe purpose of this research was to apply machine learning techniques to automate rule generation in the construction of Intelligent Tutoring Systems. By using a pair of somewhat intelligent iterative-deepening, depth-first searches, we were able to generate production rules from a set of marked examples and domain background knowledge. Such production rules required independent searches for both the "if" and "then" portion of the rule. This automated rule generation allows generalized rules with a small number of sub-operations to be generated in a reasonable amount of time, and provides non-programmer domain experts with a tool for developing Intelligent Tutoring Systems.
- Nov 4
- Zachi Klopman
Introduction to Swarm Intelligence
- Nov 11
- Terrence Turner
Designing tools to make building pseudo-intelligent tutors easierI will be discussing the challenges that arise when trying to build pseudo-intelligent tutors, and presenting on a web based software tool that allows teachers to build and edit their own pseudo-intelligent tutors, which can then be deployed with the assistment runtime environment.
- Nov 18
- Aparna Varde
Search Methods in Distance Metric LearningSearch Methods, both informed and uninformed, have been used for solving various problems. In this talk, uninformed search methods such as breadth-first and depth-first search, and informed search methods such as hill-climbing (or gradient descent) search, are explored. The use of search methods, in particular gradient descent, is considered within the context of our dissertation sub-problem. This involves learning a distance metric to accurately cluster 2-dimensional graphical plots incorporating domain semantics. Our proposed approach for distance metric learning called LearnMet is described here. This approach aims to minimize the error between actual clusters of graphs provided by domain experts and predicted clusters obtained with a clustering technique, to learn a domain-specific distance metric. This is done by performing a gradient descent search guided by heuristics based on domain knowledge. The development of good heuristics is crucial to ensure the effectiveness of LearnMet. The possibility of considering other search methods as an alternative to or in addition to gradient descent is also explored and forms a topic of our ongoing work.
- Nov 25
- Thanksgiving Break
- Dec 2
- Goss F. Nuzzo-Jones and Jason A. Walonoski
Intelligent Tutoring System Runtime ArchitectureIn this talk, we will outline some of the specialized requirements for Intelligent Tutoring System (ITS) architectures, as well as extensibile deployment demands, that had to be satisfied in the design and construction of the Tutoring Research Groups Assistments runtime environment. The presentation covers lessions learned from previous work, design decisions and rationale specific to ITS and our system in particular.
- Dec 16
- No meeting - Last day of the term/semester
AIRG Coordinator / Fall 2004