Keejun Han

MovieMine: personalized movie content search by utilizing user comments

User comments are one of the common online resources that reflect users evaluative opinions about multimedia contents. The words used in user comments provide many clues about the users and about what they like or dislike. In this paper, we propose a novel query expansion method that utilizes user comments in order to consider users different preferences in finding movies. We propose a personalized search system, called MovieMine, built upon this proposed method to provide personalized search results by expanding queries on the basis of earlier comments left by the user. Using an actual movie review dataset obtained from a large movie portal, we show that our system produces a significant performance improvement compared to the baseline condition. We expect our approach to be readily applicable to personalized searching of multimedia contents.

AppTrends: A graph-based mobile app recommendation system using usage history

With the advent of smartphones, mobile phones have evolved from a simple communication tool to a multipurpose device that affects every aspect of our daily life. The expansion of the mobile application market has made it difficult for smartphone users to find applications that fit their needs. Most prior research on application recommendation provides a limited solution to the problem of application overload. These recommendation techniques, developed outside of the mobile environment, have a number of limitations such as cold start problem and domain disparity. In this paper, we propose AppTrends, which incorporates a graph-based technique for application recommendation in the Android OS environment. Our experiment results obtained from the field usage record of over 4 million applications clearly show that the proposed graph-based recommendation model is more accurate than the Slope One Model.

Quality-Based Automatic Classification for Presentation Slides

Computerized presentation slides have become essential for effective business meetings, classroom discussions, and even general events and occasions. With the exploding number of online resources and materials, locating the slides of high quality is a daunting challenge. In this study, we present a new, comprehensive framework of information quality developed specifically for computerized presentation slides on the basis of a user study involving 60 university students from two universities and extensive coding analysis, and explore the possibility of automatically detecting the information quality of slides. Using the classifications made by human annotators as the golden standard, we compare and evaluate the performance of alternative information quality features and dimensions. The experimental results support the validity of the proposed approach in automatically assessing and classifying the information quality of slides.

Scientific table type classification in digital library

Tables are ubiquitous in digital libraries and on the Web, utilized to satisfy various types of data delivery and document formatting goals. For example, tables are widely used to present experimental results or statistical data in a condensed fashion in scientific documents. Identifying and organizing tables of different types is an absolutely necessary task for better table understanding, and data sharing and reusing. This paper has a three-fold contribution: 1) We propose Introduction, Methods, Results, and Discussion (IMRAD)-based table functional classification for scientific documents; 2) A fine-grained table taxonomy is introduced based on an extensive observation and investigation of tables in digital libraries; and 3) We investigate table characteristics and classify tables automatically based on the defined taxonomy. The preliminary experimental results show that our table taxonomy with salient features can significantly improve scientific table classification performance.

Exploiting Knowledge Structure for Proximity-aware Movie Retrieval Model

Current movie title retrieval models, such as IMDB, mainly focus on utilizing structured or semi-structured data. However, user queries for searching a movie title are often based on the movie plot, rather than its metadata. As a solution to this problem, our movie title retrieval model proposes a new way of elaborately utilizing associative relations between multiple key terms that exist in the movie plot, in order to improve search performance when users enter more than one keyword. More specifically, the proposed model exploits associative networks of key terms, called knowledge structures, derived from the movie plots. Using the search query terms entered by Amazon Mechanical Turk users as the golden standard, experiments were conducted to compare the proposed retrieval model with the extant state-of-the-art retrieval models. The experiment results show that the proposed retrieval model consistently outperforms the baseline models. The findings have practical implications for semantic search of movie titles particularly, and of online entertainment contents in general.

Introducing experiential knowledge platform: A smart decision supporter for field experts

Experiential knowledge is knowledge obtained through reflection on experience. In case of experiential knowledge within a specialized domain, this knowledge is strengthened over time as a field expert accumulates more experience in the chosen field. However, it is unfortunate that the knowledge is often confined within each individual in implicit form and it is hardly well-managed by an organization. Although there are several systems designed to acquire and exploit experiential knowledge, escalating maintenance costs pose serious challenges to their adoption and continuous use. In this paper, we propose a new knowledge-based system that acquires experiential knowledge through natural interactions with domain experts and keeps it growing by adding specialization rules, thereby reducing maintenance costs substantially. We also present the overall flow of how the acquired knowledge is processed and applied to decision supporting process, particularly in diagnosing potential diseases from blood tests.

Incorporating Distinct Opinions in Content Recommender System

As the media content industry is growing continuously, the content market has become very competitive. Various strategies such as advertising and Word-of-Mouth (WOM) have been used to draw people’s attention. It is hard for users to be completely free of others’ influences and thus to some extent their opinions become affected and biased. In the field of recommender systems, prior research on biased opinions has attempted to reduce and isolate the effects of external influences in recommendations. In this paper, we present a new measure to detect opinions that are distinct from the mainstream. This distinctness enables us to reduce biases formed by the majority and thus, to potentially increase the performance of recommendation results. To ensure robustness, we develop four new hybrid methods that are various mixtures of existing collaborative filtering (CF) methods and our new measure of Distinctness. In this way, the proposed methods can reflect the majority of opinions while considering distinct user opinions. We evaluate the methods using a real-life rating dataset with 5-fold cross validation. The experimental results clearly show that the proposed models outperform existing CF methods.

Exploiting Query Proximity and Graph Profiling Method for Tag-based Personalized Search in Folksonomy

Folksonomy data, which is derived from social tagging systems, is a useful source for understanding a users intention and interest. Using the folksonomy data, it is possible to create an accurate user profile which can be utilized to build a personalized search system. However there are limitations in some of the traditional methods such as Vector Space Model(VSM) for user profiling and similarity computation. This paper suggests a novel method with graph-based user and document profile which uses the proximity information of query terms to improve personalized search. We demonstrate the performance of the suggested method by comparing its performance with several state-of-the-art VSM based personalization models in two different folksonomy datasets. The results show that the proposed model constantly outperforms the other state-of-the-art personalization models. Furthermore, the parameter sensitivity results show that the proposed model is parameter-free in that it is not affected by the idiosyncratic nature of datasets.

Understanding the Difficulty Factors for Learning Materials: A Qualitative Study

Difficult materials overwhelm learners whereas easy materials deter advanced knowledge acquisition. Toward the goal of automatic assessment of learning materials, we conducted a laboratory experiment involving 50 college students recruited from two universities in Korea using 115 PowerPoint files. On the basis of the qualitative analysis results, we propose a model of learning difficulty, distinguishing measurable factors from non-measurable factors. The most influential factors for the easiest and the hardest learning materials are also identified and compared. The study findings have implications for educational service providers who need to automatically classify learning materials based on their innate difficulties.