• Journal of Internet Computing and Services
• /
• v.8 no.2
• /
• pp.95-103
• /
• 2007

Many techniques have been proposed for mapping from XML to relations, but most techniques did not negotiate the semantics of XML data. The semantics is important to validate storage, query optimization, modification anomaly in process of schema design. Specially, functional dependencies are an important part of database theory, also it is basis of normalization for relational table in BCNF. This paper propose a new technique that reflect functional dependencies to store relation mapped from XML based on XML Schema. The technique can reduce storage redundancy and can keep up content and structure with constraint described by functional dependencies.

• Journal of the Korea Society of Computer and Information
• /
• v.13 no.4
• /
• pp.185-193
• /
• 2008

In spite of rapid development of IT-related meteorology and services, human users still ought to check the weather information manually as they did before because traditional weather information retrieval is based on pull-type and human interpretation. Furthermore, the automatic machine-driven weather information processing has been neglected for a long time although the intelligent weather information processing is expected to be very useful for personal daily life and ubiquitous computing. In this paper, we discussed a design of GRIB based ontology to enable smart weather information processing. GRIB is the general purposed and world-wildly used weather data format approved by the World Meteorological Organization. With the designed ontology and the inference system containing Jess engine, several intelligent weather applications have been implemented and tested to verify the virtue of machine-driven weather information processing.

• Journal of KIISE:Computing Practices and Letters
• /
• v.8 no.4
• /
• pp.468-478
• /
• 2002

In this paper, we designed a B-WLL MAC(Media Access Control) protocol and validated its operation for implementation of high-speed subscriber networks. Our MAC protocol was designed by SDL using the DAVIC specifications based upon the variable contention/reserved time slot allocation algorithm. For validation of our MAC protocol, Syntax and semantic error check were performed by the Simulation Builder of ObjectGeode and the MAC(Message Sequence Chart) respectively. The validation results showed that our B-WLL MAC protocol is working correctly and may successfully support B-WLL services.

• Journal of Internet Computing and Services
• /
• v.10 no.6
• /
• pp.113-121
• /
• 2009

In order to retrieve images with different emotions in regions of the images, this paper proposes the image retrieval system using emotion sketch. The proposed retrieval system divides an image into $17{\times}17$ sub-regions and extracts emotion features in each sub-region. In order to extract the emotion features, this paper uses emotion colors on 160 emotion words from H. Nagumo's color scheme imaging chart. We calculate a histogram of each sub-region and consider one emotion word having the maximal value as a representative emotion word of the sub-region. The system demonstrates the effectiveness of the proposed emotion sketch and our experimental results show that the system successfully retrieves on the Corel image database.

• Journal of Digital Convergence
• /
• v.17 no.11
• /
• pp.517-522
• /
• 2019

In many businesses, chatbots enhance the user experience by providing the most immediate and direct feedback to user questions. The area of use of chatbots is growing. In this study, the three types of chatbot definition, command method, function, and platform are classified according to their distinct factors. In the process, the functional delimiter element is necessary for the Chatbot UX, which is a key technical element of the functional part of pattern recognition, natural language processing, semantic web, text mining, and context-aware computing. However, the limitations at this stage were also known. Based on this, we analyzed the chatbot's UX elements for Facebook, Skype, Telegram, and Google Assistant for a better user experience. Basic UI elements such as cards, quick response, command, and application of persistent menus are needed as user experience elements.

• Journal of Internet Computing and Services
• /
• v.22 no.3
• /
• pp.37-43
• /
• 2021

In this paper, we study how objects from front and back cameras can be recognized in real time in a mobile environment to segment regions of object pixels and synthesize them through image processing. To this work, we applied DeepLabV3 machine learning model to dual cameras provided by Apple's iOS. We also propose methods using Core Image and Core Graphics libraries from Apple for image synthesis and postprocessing. Furthermore, we improved CPU usage than previous works and compared the throughput rates and results of Depth and DeepLabV3. Finally, We also developed a camera application using these two methods.

• Annual Conference on Human and Language Technology
• /
• 2022.10a
• /
• pp.473-476
• /
• 2022

문장 임베딩은 문장의 의미를 고려하여 모델이 적절하게 의미적인 벡터 공간에 표상하는 것이다. 문장 임베딩을 위해 다양한 방법들이 제안되었지만, 최근 가장 높은 성능을 보이는 방법은 대조적 학습 방법이다. 대조적 학습을 이용한 문장 임베딩은 문장의 의미가 의미적으로 유사하면 가까운 공간에 배치하고, 그렇지 않으면 멀게 배치하도록 학습하는 방법이다. 이러한 대조적 학습은 비지도와 지도 학습 방법이 존재하는데, 본 논문에서는 효과적인 비지도 학습방법을 제안한다. 기존의 비지도 학습 방법은 문장 표현을 학습하는 언어모델이 자체적인 정보를 활용하여 문장의 의미를 구별한다. 그러나, 하나의 모델이 판단하는 정보로만 문장 표현을 학습하는 것은 편향적으로 학습될 수 있기 때문에 한계가 존재한다. 따라서 본 논문에서는 Cross-Encoder의 Re-Ranker를 통한 의미 검색으로부터 추천된 문장 쌍을 학습하여 기존 모델의 성능을 개선한다. 결과적으로, STS 테스크에서 베이스라인보다 2% 정도 더 높은 성능을 보여준다.

• Journal of KIISE:Software and Applications
• /
• v.34 no.9
• /
• pp.817-827
• /
• 2007

In the age of ubiquitous environment, huge number of devices and computing services are provided to users. Personalized service, which is modeled according to the character of each and every individual is of particular need. In order to provide various dynamic services according to user's movement, service unit and operating mode should be able to operate automatically with minimum user intervention. In this paper, we discuss the steps of offering approximate service based on user's request in ubiquitous environment. First, we present our simulator designed for modeling the physical resource and computing object in smart space - the infrastructure in ubiquitous. Second, intelligent agents, which we developed based on a FIPA specification compliant multi-agent framework will be discussed. These intelligent agents are developed for achieving the service goal through cooperation between distributed agents. Third, we propose an automated service discovery and composition method in heterogeneous environment using semantic message communication between agents, according to the movement by the user interacting with the service available in the smart space. Fourth, we provide personalized service through agent monitoring anytime, anywhere from user's profile information stored on handhold device. Therefore, our research provides high quality service more than general automated service operation.

• Journal of KIISE:Databases
• /
• v.27 no.1
• /
• pp.53-63
• /
• 2000

The mobile computing service market is increasing rapidly due to the development of low-cost wireless network technology and the high-performance mobile computing devices. In recent years, several methods have been proposed to effectively deal with restrictions of the mobile computing environment such as limited bandwidth, frequent disconnection and short-lived batteries. Amongst those methods, much study is being done on the caching method - among the data transmitted from a mobile support station, it selects those that are likely to be accessed in the near future and stores them in the local cache of a mobile host. Existing cache replacement methods have some limitations in efficiency because they do not take into consideration the characteristics of user mobility and spatial attributes of geographical data. In this paper, we show that the value and the semantic of the data, which are stored in the cache of a mobile host, changes according to the movement of the mobile host. We argue it is because data that are geographically near are better suited to provide an answer to a users query in the mobile environment. Also, we define spatial location of geographical data has effect on, using the spatial attributes of data. Finally, we propose two new cache replacement methods that efficiently support user mobility and spatial attributes of data. One is based on the location of data and the other on the meaningful region of data. From the comparative analysis of the previous methods and that they improve the cache hit ratio. Also we show that performance varies according to data density using this, we argue different cache replacement methods are required for regions with varying density of data.

• Journal of KIISE:Software and Applications
• /
• v.34 no.12
• /
• pp.1045-1055
• /
• 2007

Latent topic models are statistical models which automatically captures salient patterns or correlation among features underlying a data collection in a probabilistic way. They are gaining an increased popularity as an effective tool in the application of automatic semantic feature extraction from text corpus, multimedia data analysis including image data, and bioinformatics. Among the important issues for the effectiveness in the application of latent topic models to the massive data set is the efficient learning of the model. The paper proposes an accelerated learning technique for PLSA model, one of the popular latent topic models, by an incremental EM algorithm instead of conventional EM algorithm. The incremental EM algorithm can be characterized by the employment of a series of partial E-steps that are performed on the corresponding subsets of the entire data collection, unlike in the conventional EM algorithm where one batch E-step is done for the whole data set. By the replacement of a single batch E-M step with a series of partial E-steps and M-steps, the inference result for the previous data subset can be directly reflected to the next inference process, which can enhance the learning speed for the entire data set. The algorithm is advantageous also in that it is guaranteed to converge to a local maximum solution and can be easily implemented just with slight modification of the existing algorithm based on the conventional EM. We present the basic application of the incremental EM algorithm to the learning of PLSA and empirically evaluate the acceleration performance with several possible data partitioning methods for the practical application. The experimental results on a real-world news data set show that the proposed approach can accomplish a meaningful enhancement of the convergence rate in the learning of latent topic model. Additionally, we present an interesting result which supports a possible synergistic effect of the combination of incremental EM algorithm with parallel computing.