Yi-Shin Chen

Yoda: An Accurate and Scalable Web-Based Recommendation System

Recommendation systems are applied to personalize and customize the Web environment. We have developed a recommendation system, termed Yoda, that is designed to support large-scale Web-based applications requiring highly accurate recommendations in real-time. With Yoda, we introduce a hybrid approach that combines collaborative filtering (CF) and content-based querying to achieve higher accuracy. Yoda is structured as a tunable model that is trained off-line and employed for real-time recommendation on-line. The on-line process benefits from an optimized aggregation function with low complexity that allows realtime weighted aggregation of the soft classification of active users to predefined recommendation sets. Leveraging on localized distribution of the recommendable items, the same aggregation function is further optimized for the off-line process to reduce the time complexity of constructing the pre-defined recommendation sets of the model. To make the off-line process scalable furthermore, we also propose a filtering mechanism, FLSH, that extends the Locality Sensitive Hashing technique by incorporating a novel distance measure that satisfies specific requirements of our application. Our end-to-end experiments show while Yodas complexity is low and remains constant as the number of users and/or items grow, its accuracy surpasses that of the basic nearest-neighbor method by a wide margin (in most cases more than 100%).

Web Information Personalization: Challenges and Approaches

As the number of web pages increases dramatically, the problem of the information overload becomes more severe when browsing and searching the WWW. To alleviate this problem, personalization becomes a popular remedy to customize the Web environment towards a user’s preference. To date, recommendation systems and personalized web search systems are the most successful examples of Web personalization. By focusing on these two types of systems, this paper reviews the challenges and the corresponding approaches proposed in the past ten years.

An Artificial Neural-Network-Based Approach to Software Reliability Assessment

In this paper, we propose an artificial neural- network-based approach for software reliability estimation and modeling. We first explain the network networks from the mathematical viewpoints of software reliability modeling. That is, we will show how to apply neural network to predict software reliability by designing different elements of neural networks. Furthermore, we will use the neural network approach to build a dynamic weighted combinational model. The applicability of proposed model is demonstrated through four real software failure data sets. From experimental results, we can see that the proposed model significantly outperforms the traditional software reliability models.

TweoLocator: a non-intrusive geographical locator system for Twitter

In the last decade, the Internet has seen the rise of social networking as the number one online activity worldwide. To estimate the geographical location of users of social networks at a particular moment, we propose an approach to geo-tag Twitter users based only on the content of their posts. These data can later be used for local sentiment analysis, emergency detection, finding a missing person, and other novel location-based purposes. Our approach carries out a semantic analysis of tweet content to infer where in the globe a particular user is located at a given time. Based on our experimental results, conducted through Amazon Mechanical Turk, the proposed framework was evaluated by 93 evaluators who assessed 654 twitter user profiles and 2,165 tweets from 17 countries. Our system inferred some geographical information for 81% of evaluated profiles. Results show 79% accuracy in identifying the users country and 66% accuracy in identifying the users current location. This high accuracy shows that our proposed method is feasible and effective.

Soft query in image retrieval systems

We explore the use of soft computing and user defined classifications in multimedia database systems for content- based queries to obtain the members of a class is a fixed set. With multimedia databases, however, an object may belong to different classes with different probabilities. In addition, alternative users may classify objects differently due to subjectivity of human perception on multimedia objects. In order to remedy for this situation, we propose a unified model that captures both conventional techniques and soft memberships. We implemented the model by extending the traditional database query capabilities such that the result of a query depends on the user who submits the query. We compared our proposed system with conventional image retrieval systems and observed a significant margin of improvement in matching the user expectations.

Detect phishing by checking content consistency

Phishing is a form of cybercrime used to lure a victim to reveal his/her sensitive personal information to fraudulent web pages. To protect users from phishing attacks, many anti-phishing techniques have been proposed to block suspicious web pages, which are identified against registered black-lists, or checked by search engines. However, such approaches usually have difficulty in keeping up with the rapidly emerging phishing web pages. To lessen this problem, this paper proposes a technique for identifying suspicious web pages, based on the literal and conceptual consistency between the URL and web contents. By using the search logs only as reference data, our approach can achieve 98% accuracy, showing that it is effective in detecting various forms of phishing attack.

An interactive conducting system using Kinect

A conducting system for conductors of all skill levels is in high demand. An interactive conducting system could mediate the interpretation of a conductors gestures and the performance of music based on recognized results from each distributed player. Such systems would be a helpful training tool for students, an experiencing tool for professional conductors and composers to shape music at a low cost, or an entertainment tool for nonprofessional music lovers. In this paper, we propose a real-time interactive conducting system using Microsoft Kinect. The proposed system overcomes the limitation that Kinect is usually designed for large body movements; hence, delicate conducting signals can be correctly recognized without referencing any prior knowledge. The system was evaluated by conductors of all skill levels and had a high level of accuracy and a low latency.

Using Ontology to Map Categories in Blog

This paper proposes a framework to automatically map user-defined categories in blog. The proposed framework is composed of a series of procedures including information extraction from the blog, building the personal ontology, and comparing semantic similarities between user-defined categories. Our novel semantic similarity techniques can determine how similar two sets of information concepts are, based on a given ontology. The experimental results demonstrate that our framework and our proposed semantic similarity techniques are effective

Event identification for social streams using keyword-based evolving graph sequences

Social networks, which have become extremely popular nowadays, contain a tremendous amount of user-generated content about real-world events. This user-generated content can naturally reflect the real-world event as they happen, and sometimes even ahead of the newswire. The goal of this work is to identify events from social streams. A model called “keyword-based evolving graph sequences” (kEGS) is proposed to capture the characteristics of information propagation in social streams. The experimental results show the usefulness of our approach in identifying real-world events in social streams.

Interacting with a Music Conducting System

Music conducting is the art of directing musical ensembles with hand gestures to personalize and diversify a musical piece. The ability to successfully perform a musical piece demands intense training and coordination from the conductor, but preparing a practice session is an expensive and time-consuming task. Accordingly, there is a need for alternatives to provide adequate training to conductors at all skill levels; virtual reality technology holds promise for this application. The goal of this research was to study the mechanics of music conducting and develop a system capable of closely simulating the conducting experience. After extensive discussions with professional and nonprofessional conductors, as well as extensive research on music conducting material, we identified several key features of conducting. A set of lightweight algorithms exploring those features were developed to enable tempo control and instrument emphasis, two core components of conducting. By using position/orientation sensors and data gloves as the interface for human-computer interaction, we developed a functional version of the system. Evaluating the algorithms in real-world scenarios gave us promising results; most users of the final system expressed satisfaction with the virtual experience.