• Journal of KIISE
• /
• v.43 no.8
• /
• pp.902-909
• /
• 2016

Question answering system (QA system) is a system that finds an actual answer to the question posed by a user, whereas a typical search engine would only find the links to the relevant documents. Recent works related to the open domain QA systems are receiving much attention in the fields of natural language processing, artificial intelligence, and data mining. However, the prior works on QA systems simply replace all words that are not in the training data with a single token, even though such unseen words are likely to play crucial roles in differentiating the candidate answers from the actual answers. In this paper, we propose a method to compute vectors of such unseen words by taking into account the context in which the words have occurred. Next, we also propose a model which utilizes inverse document frequencies (IDF) to efficiently process unseen words by expanding the system's vocabulary. Finally, we validate that the proposed method and model improve the performance of a QA system through experiments.

• Journal of Internet Computing and Services
• /
• v.9 no.1
• /
• pp.43-54
• /
• 2008

In this paper, we implement an automatic entity information management system for smart refrigerator using RFID technology in which objects containing electronic tags are automatically identified using radio wave. Unlike current "smart" refrigerators, the system presented in this paper implements smart tag information acquisition mechanism and real-time information management system to provide various information on entities in refrigerators to local and remote users. As the first step, this paper analyzes the requirements for smart refrigerator system based on the RFID and suggests design considerations. Based on the analysis, we propose and implement an efficient tag location tracking method based on antenna transfer method and an intelligent tag information management system based on embedded database and web server. We also provide a wide range of experimental results on the number of tags identified at a time and the tag recognition ratio according to the RFID antenna transfer speed and the angle between tag reader and tags.

• Journal of Digital Contents Society
• /
• v.18 no.4
• /
• pp.739-745
• /
• 2017

This paper presents and discusses an implementation of the GPU shifted sorting method to find approximate k nearest neighbors which executes within "warp", the minimum execution unit in GPU parallel architecture. Also, this paper presents the comparison results with other two common nearest neighbor searching methods, GPU-based kd-tree and ANN (Approximate Nearest Neighbor) library. The proposed implementation focuses on the cases when k is small, i.e. 2, 4, 8, and 16, which are handled efficiently within warp to consider it is very common for applications to handle small k's. Also, this paper discusses optimization ways to implementation by improving memory management in a loop for the CUB open library and adopting CUDA commands which are supported by GPU hardware. The proposed implementation shows more than 16-fold speed-up against GPU-based other methods in the tests, implying that the improvement would become higher for more larger input data.

• Journal of the Korean Institute of Telematics and Electronics S
• /
• v.36S no.10
• /
• pp.113-122
• /
• 1999

When we want to retrieve the most similar image from the image database, the color histogram intersection, shape feature and texture feature comparing method are used as a metric to measure the similarity. In order to increase the accuracy of retrievals, we need to integrate two different features. In this paper, the histogram intersection and shape based block histogram intersection method are used. This method results in a high efficient algorithm that meets a similar accuracy and a relatively fast retrieval speed compared to the method of integration of two different features. The Proposed algorithm is tested on retrievals of image database consisting of various 600 images and we implemented that the proposed algorithm gives fast, high efficiency and reliability compared to others.

• Journal of Digital Convergence
• /
• v.13 no.9
• /
• pp.235-242
• /
• 2015

Social network technology has been increasing interest in the big data service and development. However, the data stored in the distributed server and not on the central server technology is easy enough to find and extract. In this paper, we propose a big data management techniques to minimize the processing time of information you want from the content server and the management server that provides big data services. The proposed method is to link the in-group data, classified data and groups according to the type, feature, characteristic of big data and the attribute information applied to a hash chain. Further, the data generated to extract the stored data in the distributed server to record time for improving the data index information processing speed of the data classification of the multi-attribute information imparted to the data. As experimental result, The average seek time of the data through the number of cluster groups was increased an average of 14.6% and the data processing time through the number of keywords was reduced an average of 13%.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.14 no.1
• /
• pp.63-68
• /
• 2004

In this paper, we propose a method about the extraction of vehicle and tracking trajectory for moving vehicle tracking system in road. This system applied to the monitoring system of the traffic flow for ATMS(advanced traffic management system) of ITS(intelligent transport system). Also, this system can solve the problem of maintenance of loop sensor. And we detected vehicle using differential image analysis. Because of the road environment changes by real time. Therefore, the method to use background image is not suitable. And we used Kalman filter and innovation value and variable search area for vehicle tracking system. Previous method using fixed search area is sensitive to the moving trajectory and the speed of vehicle. Simulation results show that proposed method increases the possibility of traffic measurement more than fixed area traffic measurement system.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.18 no.9
• /
• pp.2209-2215
• /
• 2014

This paper presents, a deinterlacing method by improving the Direction-Oriented Interpolation (DOI) technique. The technique is considered to be a very strong tool for intrafield-based deinterlacing. However, DOI has some problems such as long processing time, wrong edge detection in periodic pattern. To remedy this problem, we replace the full search in DOI by a two-step search to reduce processing time and introduces two additional processes to improve image quality. In the proposed method, the spatial direction vectors (SDVs) misread data are reconsidered to prevent them utilizing in the next interpolation step, resulting in an accurate deinterlacing method. We conduct experiments with ISO experimental images to compare the proposed method with the existing methods including line evarage (LA), edge-based line averaging (ELA), DOI, selective deinterlacing algorithm (SDA). Experimental results show the proposed method gives better performance in objective and subjective quality than existing deinterlacing methods.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.21 no.2
• /
• pp.333-338
• /
• 2017

Recently, because of the mobile device generalization tendency, a lot of new mobile applications which need mobile database on mobile devices appear. SQLite, Realm, etc. are representative mobile databases, and they are the database engines which are suitable for small devices that have large resource restriction and are used a lot in mobile devices based on Linux. These engines are open source programs and have advantages which are lighter weight and faster speed than general purpose databases. This paper focuses on the performance experiments and analysis of SQLite, which is front runner in mobile database fields. Considering the target mobile environments, the test is performed at Linux environment. And this paper implements performance test program and performs performance experiment of SQLite database to test search efficiency. The experiment is processed by comparison with the results of a Oracle database working out at the same environment.

• Journal of KIISE:Databases
• /
• v.32 no.6
• /
• pp.650-661
• /
• 2005

As the Internet is growing, the use of XML which is a standard of semi-structured document is increasing. Therefore, there are on going works about integration and retrieval of XML documents. However, the basis of efficient integration and retrieval of documents is to cluster XML documents with similar structure. The conventional XML clustering approaches use the hierarchical clustering algorithm that produces the demanded number of clusters through repeated merge, but it have some problems that it is difficult to compute the similarity between XML documents and it costs much time to compare similarity repeatedly. In order to address this problem, we use clustering algorithm for transactional data that is scale for large size of data. In this paper we use common structures from XML documents that don't have DTD or schema. In order to use common structures of XML document, we extract representative structures by decomposing the structure from a tree model expressing the XML document, and we perform clustering with the extracted structure. Besides, we show efficiency of proposed method by comparing and analyzing with the previous method.

• Journal of KIISE:Computer Systems and Theory
• /
• v.30 no.10
• /
• pp.586-594
• /
• 2003

X-Match algorithm is a lossless compression algorithm suitable for hardware implementation owing to its simplicity. It can compress 32 bits per clock cycle and is suitable for real time compression. However, as the bus width increases 64-bit, the compression unit also need to increase. This paper proposes the cooperative parallel X-Match (X-MatchCP) algorithm, which improves the compression speed by performing the two X-Match algorithms in parallel. It searches the all dictionary for two words, combines the compression codes of two words generated by parallel X-Match compression and outputs the combined code while the previous parallel X-Match algorithm searches an individual dictionary. The compression ratio in X-MatchCP is almost the same as in X-Match. X-MatchCP algorithm is described and simulated by Verilog hardware description language.