| Thongtanunam Patanamon | 1251211: M, 1回目発表 |
|
title: A study of expertise finding in Q&A community based on award clustering using sequence analysis
abstract: Finding an expertise in Q&A online communities would reduce development and maintenance cost by helping developers to solve problems faster. Currently, Q&A online communities adopt an award strategy to engage aspiring users and this award achievement directly correspond to user’s activities. Therefore, this award achievement history would be a meaningful information for finding experts. In this presentation, I introduce my current research on how to mine award achievement history of the StackOverflow for finding experts on specific activities. I used sequence analysis and clustering techniques for such a purpose. To the best of my knowledge, this is the first study of award achievement history to find expertise. Using this novel approach, I effectively discriminated users and uncovered meaningful clusters and their key activities. From the resulting clusters, user candidates to verify an answer and to improve questions can be selected more confidently. language of the presentation: English | |
| Sornchumni Nut | 1251210: M, 1回目発表 |
|
title: Concept and application of Differential Privacy
abstract: Privacy is one of the main concerns in database query processing and data mining. Unfortunately, it has been mathematically proved that perfect privacy cannot be achieved if auxiliary information is available. Differential privacy is a new measure which captures the increased risk to one’s privacy incurred by participating in a database. In this presentation, I will first explain the background on database privacy, followed by the definition of differential privacy. Also I will introduce an existing study on data mining with differential privacy. language of the presentation: English | |
| 万 耕 | 1251214: M, 1回目発表 |
|
title: Performance Evaluation of STBC for Dual-polarized MIMO-OFDM Systems
abstract: Multiple-Input Multiple-Output (MIMO) Orthogonal Frequency Division Multiplexing (OFDM) wireless systems can provide high-speed data transmission over broadband wireless channels. However, since the requirements for antenna spacing will lead to high im plementation cost, the use of dual-polarized antennas appears as an effective alternative. In this research, the channel model of 4x4 dual-polarized MIMO systems based on GSCM (Geometry-Based Stochastic Channel Model) is firstly studied. Then, the BER performance of STBC (Space-Time Block Codes) for conventional dual-polarized MIMO systems is compared with that of uni-polarized MIMO systems. However, due to the impact of correlations between dual-polarized arrays, the BER performance of STBC for conventional dual-polarized MIMO systems cannot achieve that of uni-polarized MIMO systems. The validity of conclusion above is verified by the simulation result. language of the presentation: English | |
| Tantithamthavorn Chakkrit | 1251205: M, 1回目発表 |
| Title: Toward Practical Bug Localization: Finding Buggy Methods Accurately and Effectively Abstract: Bug localization can help developers find a bug location from billions of instructions. There is much research on file-level bug localization techniques. However, these techniques often leave developers with a huge amount of effort to examine all instructions in a file until a bug is located, which is not practical for developers. In this research, we address a practical bug localization technique. The aim of this technique is to accurately and effectively identify locations of bugs. We use the strength of Vector Space Model (VSM) and consider at method level of granularity. We measure the accuracy and effectiveness of bug localization using effort-based evaluation. We expect that method-level technique will identify bug locations more accurately and effectively than file-level technique, which can save human effort to locate bugs. Language of the presentation: English | |