Dong-Kyu Chae

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.

Effective and efficient detection of software theft via dynamic API authority vectors

We design a novel feature of a program for detecting software theft.We reflect the sequence and the frequency information of a program to our feature.Our proposed method is credible, resilient, and scalable.Our method outperforms existing software theft detection methods in our experiments. Software theft has become a very serious threat to both the software industry and individual software developers. A software birthmark indicates unique characteristics of a program in question, which can be used for analyzing the similarity of a pair of programs and detecting theft. This paper proposes a novel birthmark, a dynamic API authority vector (DAAV). DAAV satisfies four essential requirements for good birthmarkscredibility, resiliency, scalability, and packing-freewhile existing static birthmarks are unable to handle the packed programs and existing dynamic birthmarks do not satisfy credibility and resiliency. Through our extensive experiments with a set of Windows applications, DAAV is shown to have not only the credibility and resiliency higher than the existing dynamic birthmarks but also the accuracy comparable to that of existing static birthmarks. This result indicates that our proposed birthmark provides high accuracy and also covers packed programs successfully in detecting software theft.

On identifying k-nearest neighbors in neighborhood models for efficient and effective collaborative filtering

Abstract Neighborhood models (NBMs) are the methods widely used for collaborative filtering in recommender systems. Given a target user and a target item, NBMs find k most similar users or items (i.e., k-nearest neighbors) and make a prediction of a target user on an item based on the rating patterns of those neighbors on the item. In NBMs, however, we have a difficulty in satisfying both the performance and accuracy together. In order to pursue an accurate recommendation, NBMs may find the k-nearest neighbors at every recommendation request to exploit the latest ratings, which requires a huge amount of computation time. Alternatively, NBMs may search for the k-nearest neighbors offline, which consequently results in inaccurate recommendation as time goes by, or even may not able to deal with new users or new items, because they cannot exploit the latest ratings generated after the k-nearest neighbors are determined. In this paper, we propose a novel approach that finds the k-nearest neighbors efficiently by identifying only those users and items necessary in computing the similarity. The proposed approach enables NBMs not to require any offline similarity computations but to exploit the latest ratings, thereby resolving speed-accuracy tradeoff successfully. We demonstrate the effectiveness of the proposed approach through extensive experiments.

Influence maximisation in social networks: A target-oriented estimation:

Influence maximisation (IM) is the problem of finding a set of k-seed nodes that could maximize the amount of influence spread in a social network. In this article, we point out that the existing methods are taking the source-oriented estimation (SOE), which is the main reason of their failure in accurately estimating the amount of potential influence spread of an individual node. We propose a novel target-oriented estimation (TOE) that understands information diffusion more accurately as well as remedies the drawback of the existing methods. Our extensive experiments on four real-world datasets demonstrate that our proposed method outperforms the existing methods consistently with respect to the quality of the selected seed set.

A Single-Step Approach to Recommendation Diversification

This paper addresses recommendation diversification. Existing diversification methods have a difficulty in dealing with the accuracy-diversity tradeoff. We propose a novel method to simultaneously optimize the user preference and diversity of k-items to be recommended.

Accurate Path-based Methods for Influence Maximization in Social Networks

This paper proposes a novel approach to target-oriented influence estimation, which remedies the drawback of state-of-the-art, thereby understanding information diffusion more accurately in a social network.

CFGAN: A Generic Collaborative Filtering Framework based on Generative Adversarial Networks

Generative Adversarial Networks (GAN) have achieved big success in various domains such as image generation, music generation, and natural language generation. In this paper, we propose a novel GAN-based collaborative filtering (CF) framework to provide higher accuracy in recommendation. We first identify a fundamental problem of existing GAN-based methods in CF and highlight it quantitatively via a series of experiments. Next, we suggest a new direction of vector-wise adversarial training to solve the problem and propose our GAN-based CF framework, called CFGAN, based on the direction. We identify a unique challenge that arises when vector-wise adversarial training is employed in CF. We then propose three CF methods realized on top of our CFGAN that are able to address the challenge. Finally, via extensive experiments on real-world datasets, we validate that vector-wise adversarial training employed in CFGAN is really effective to solve the problem of existing GAN-based CF methods. Furthermore, we demonstrate that our proposed CF methods on CFGAN provide recommendation accuracy consistently and universally higher than those of the state-of-the-art recommenders.

Crowdsourced promotions in doubt: Analyzing effective crowdsourced promotions

Abstract Recently, crowdsourcing systems have been adopted for promoting products in online social networks (OSN), e.g. , Twitter. We call it the crowdsourced promotion . When promoting products using crowdsourcing systems, it is critical to qualify the effectiveness of such promotions in OSN. One possible solution is to use conventional attributes for the characteristics of workers such as worker levels, the number of followers , and Klout scores . Unlike existing crowdsourcing tasks that are performed in crowdsourcing systems, crowdsourced promotions are mainly performed in OSN. Therefore, conventional attributes for workers are ineffective for validating the quality of crowdsourced promotions. In this paper, we propose a new method for measuring the effectiveness of crowdsourced promotions. It is important to determine whether workers can deliver promotional messages to legitimate users in OSN. In other words, because workers usually propagate the promotional messages to their followers, we aim to measure the ratio of legitimate users to the followers of the worker. Toward this goal, we first devise various attributes to identify legitimate users among all followers. Then, using these attributes, we build a classifier to distinguish between legitimate and non-legitimate users. Lastly, we measure the effectiveness of crowdsourced promotions by using the ratio of legitimate users to followers. Our empirical study demonstrates that the proposed method outperforms the existing baseline methods using conventional attributes.

On Classifying Dynamic Graph Bags

In this paper, we introduce a novel problem of dynamic graph bag classification, and propose a method to solve this problem. Here, a graph bag (simply, bag) corresponds to a training object that contains one or multiple graphs. Dynamic bag classification aims to build a classification model for bags which are presented in a dynamic fashion, i.e., emerging of new bags or graphs. Our proposed solution for this problem can gradually update the classification model whenever such changes are made to a bag dataset, rather than building a model from the scratch. We demonstrate the effectiveness of our proposed method by our extensive evaluation on a real-world graph dataset.