Jiwoon Ha

Top-N recommendation through belief propagation

The top-n recommendation focuses on finding the top-n items that the target user is likely to purchase rather than predicting his/her ratings on individual items. In this paper, we propose a novel method that provides top-n recommendation by probabilistically determining the target users preference on items. This method models the purchasing relationships between users and items as a bipartite graph and employs Belief Propagation to compute the preference of the target user on items. We analyze the proposed method in detail by examining the changes in recommendation accuracy under different parameter settings. We also show that the proposed method is up to 40% more accurate than an existing method by comparing it with an RWR-based method via extensive experiments.

Software plagiarism detection: a graph-based approach

As plagiarism of software increases rapidly, there are growing needs for software plagiarism detection systems. In this paper, we propose a software plagiarism detection system using an API-labeled control flow graph (A-CFG) that abstracts the functionalities of a program. The A-CFG can reflect both the sequence and the frequency of APIs, while previous work rarely considers both of them together. To perform a scalable comparison of a pair of A-CFGs, we use random walk with restart (RWR) that computes an importance score for each node in a graph. By the RWR, we can generate a single score vector for an A-CFG and can also compare A-CFGs by comparing their score vectors. Extensive evaluations on a set of Windows applications demonstrate the effectiveness and the scalability of our proposed system compared with existing methods.

Software plagiarism detection via the static API call frequency birthmark

In this paper, we propose a system for detecting software plagiarism using a birthmark. The birthmark is representative features of a program, which can be used to identify the program. We use a set of frequency of APIs used in a program as its birthmark. The proposed system consists of three components. First, it extracts the frequency of APIs employed in a program. Next, it generates the program birthmark using a set of frequency of APIs and weights to APIs to extract unique features of the program. Finally, it decides the plagiarism based on the cosine similarity between the birthmarks. Through extensive experiments, it was found that the proposed system can provide 97.2% of precision and 95.7% of recall in plagiarism detection.

Recommendation of newly published research papers using belief propagation

The problem to retrieve most relevant research papers for a given academic is studied. Existing solutions cannot adequately address the recommendation of new papers due to their lack of history information, the so-called cold start problem. Using the graphical model built from citation information between a new paper pi and published papers, toward this challenge, we propose a novel approach based on a probabilistic inference algorithm, the Belief Propagation (BP), to predict the likelihood of pis relevance to a target academic. Compared to item-based collaborative filtering method using a DBLP data set, the empirical validation shows an improvement in accuracy up to 26% in F1 score.

Link-based similarity measures using reachability vectors.

We present a novel approach for computing link-based similarities among objects accurately by utilizing the link information pertaining to the objects involved. We discuss the problems with previous link-based similarity measures and propose a novel approach for computing link based similarities that does not suffer from these problems. In the proposed approach each target object is represented by a vector. Each element of the vector corresponds to all the objects in the given data, and the value of each element denotes the weight for the corresponding object. As for this weight value, we propose to utilize the probability of reaching from the target object to the specific object, computed using the “Random Walk with Restart” strategy. Then, we define the similarity between two objects as the cosine similarity of the two vectors. In this paper, we provide examples to show that our approach does not suffer from the aforementioned problems. We also evaluate the performance of the proposed methods in comparison with existing link-based measures, qualitatively and quantitatively, with respect to two kinds of data sets, scientific papers and Web documents. Our experimental results indicate that the proposed methods significantly outperform the existing measures.

Credible, resilient, and scalable detection of software plagiarism using authority histograms

Software plagiarism has become a serious threat to the health of software industry. A software birthmark indicates unique characteristics of a program that can be used to analyze the similarity between two programs and provide proof of plagiarism. In this paper, we propose a novel birthmark, Authority Histograms (AH), which can satisfy three essential requirements for good birthmarks-resiliency, credibility, and scalability. Existing birthmarks fail to satisfy all of them simultaneously. AH reflects not only the frequency of APIs, but also their call orders, whereas previous birthmarks rarely consider them together. This property provides more accurate plagiarism detection, making our birthmark more resilient and credible than previously proposed birthmarks. By random walk with restart when generating AH, we make our proposal fully applicable to even large programs. Extensive experiments with a set of Windows applications verify that both the credibility and resiliency of AH exceed those of existing birthmarks; therefore AH provides improved accuracy in detecting plagiarism. Moreover, the construction and comparison phases of AH are established within a reasonable time.

BlogCast effect on information diffusion in a blogosphere

A blog service company provides a function named BlogCast that exposes quality posts on the blog main page to vitalize a blogosphere. This paper analyzes a new type of information diffusion via BlogCast. We show that there exists a strong halo effect in a blogosphere via thorough investigation on a huge volume of blog data.

Analyzing a Korean blogosphere: a social network analysis perspective

Due to their popularity and widespread use, blogs have become an important medium through which to communicate and exchange information on the World Wide Web. The advent of the blogosphere may provide opportunities for establishing a new business model that investigates social relationships. In Korea, there are many blogospheres that appear to maintain different characteristics from foreign blogospheres on the Internet. Consequently, it is inappropriate to apply analysis methods used for the foreign blogosphere directly to Korean blogospheres. To establish successful business policies in Korean blogospheres, the characteristics of Korean blogospheres and the behavioral patterns of bloggers should be understood. In this paper, we analyze the characteristics of the Korean blog network, wherein each blogger forms a node and scraps by bloggers as edges. First, we demonstrate that the Korean blog network is also a scale-free network, like the World Wide Web. Second, we compare the Bow-tie structure of the Korean blog network with that of the World Wide Web. We expect that these analysis results will be helpful in developing effective algorithms and in establishing new business models targeted at the Korean blogosphere.

Recommendation of research papers in DBpia: A Hybrid approach exploiting content and collaborative data

DBpia is the largest digital-bibliography service provider in Korea. It provides several convenience functions for researchers. DBpia users (i.e., researchers) can search for papers via several search routes such as publications, publishers, authors, and keywords. Although the researchers can exploit the search functions, they may still have a number of search results as candidate papers to read. Therefore, it is crucial to provide a function of recommending most relevant papers to an individual user. In this paper, we (1) discuss several methods with four datasets of DBpia in the context of paper recommendation using content-based or graph-based recommendation, and (2) propose a hybrid approach suitable for paper recommendation combining the content-based and the graph-based approaches. We lastly conduct extensive experiments by a real-world academic literature dataset in DBpia to verify the effectiveness of our proposed approach.

An efficient and effective method to find uninteresting items for accurate collaborative filtering

Collaborative filtering methods suffer from a data sparsity problem, which indicates that the accuracy of recommendation decreases when the user-item matrix used in recommendation is sparse. To alleviate the data sparsity problem, researches on data imputation have been done. In particular, the zero-injection method, which finds uninteresting items and imputes zero values to those items for collaborative filtering, achieves significant improvement in terms of recommendation accuracy. However, the existing zero-injection method employs the One-Class Collaborative Filtering (OCCF) method that requires a lot of time. In this paper, we propose a fast method that finds uninteresting items rapidly with preserving high recommendation accuracy. Our experimental results show that our method is faster than the existing zero-injection method and also show that the recommendation accuracy using our method is slightly higher than or similar to that of the existing zero-injection method.