• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.8
• /
• pp.3168-3186
• /
• 2020

Automatic topic essay generation (TEG) is a controllable text generation task that aims to generate informative, diverse, and topic-consistent essays based on multiple topics. To make the generated essays of high quality, a reasonable method should consider both diversity and topic-consistency. Another essential issue is the intrinsic link of the topics, which contributes to making the essays closely surround the semantics of provided topics. However, it remains challenging for TEG to fill the semantic gap between source topic words and target output, and a more powerful model is needed to capture the semantics of given topics. To this end, we propose a pretraining-based contextual self-attention (PC-SAN) model that is built upon the seq2seq framework. For the encoder of our model, we employ a dynamic weight sum of layers from BERT to fully utilize the semantics of topics, which is of great help to fill the gap and improve the quality of the generated essays. In the decoding phase, we also transform the target-side contextual history information into the query layers to alleviate the lack of context in typical self-attention networks (SANs). Experimental results on large-scale paragraph-level Chinese corpora verify that our model is capable of generating diverse, topic-consistent text and essentially makes improvements as compare to strong baselines. Furthermore, extensive analysis validates the effectiveness of contextual embeddings from BERT and contextual history information in SANs.

• Journal of KIISE
• /
• v.43 no.2
• /
• pp.237-245
• /
• 2016

In this paper we introduce a novel database research area called science of databases (SoDB) and carry out a comprehensive analysis of its case studies. SoDB aims to better understand interesting phenomena observed across multiple database management systems (DBMSes). While mathematical and engineering work in the database field has been dominant, less attention has been given to scientific approaches through which DBMSes can be better understood. Scientific investigations can lead to better engineered designs through deeper understanding of query optimizers and transaction processing. The SoDB research has investigated several interesting phenomena observed across different DBMSes and provided several engineering implications based on our uncovered results. In this paper we introduce a novel scientific, empirical methodology and describe the research infrastructure to enable the methodology. We then review each of a selected group of phenomena studied and present an identified structural causal model associated with each phenomenon. We also conduct a comprehensive analysis on the case studies. Finally, we suggest future directions to expand the SoDB research.

• Journal of the Korea Society of Computer and Information
• /
• v.13 no.4
• /
• pp.141-150
• /
• 2008

In this mobile computing environment, data broadcasting is widely used to resolve the problem of limited power and bandwidth of mobile equipments. Most previous broadcast indexing methods concentrate on flat data. However. with the growing popularity of XML, an increasing amount of information is being stored and exchanged in the XML format. We propose a novel indexing method. called TOP tree(Tree Ordering based Path summary tree), for indexing XML document on mobile broadcast environments. TOP tree is a path summary tree which provides a concise structure summary at group level using global IDs and element information at local level using local IDs. Based on the TOP tree representation, we suggest a broadcast stream generation and query Processing method that efficiently handles not only simple Path queries but also multiple path queries. We have compared our indexing method with other indexing methods. Evaluation results show that our approaches can effectively improve the access time and tune-in time in a wireless broadcasting environment.

• Journal of KIISE:Software and Applications
• /
• v.33 no.8
• /
• pp.703-717
• /
• 2006

The Semantic Web has the objective of developing universal media in which machine-processable semantic information can be represented and shared, and it is therefore important to distribute ontologies that represent this kind of semantic information to the Web and make them available to multiple parties. However, the current ontology authoring tools are not operating on the Web, which makes it difficult to distribute ontologies directly to the Web and to create and edit them collaboratively with other people. This paper proposes a framework that facilitates the ontology construction and sharing, realizing easy distribution of ontologies to the Web. Wiki is one of the frameworks for collaborative construction and sharing of knowledge on the Web, and Wiki contents consist of natural language texts and simple markup language for visualization. For better collaboration in creating and sharing ontologies, this paper suggests the Semantic Wiki that embodies the Semantic Web features to the existing Wiki system. The Semantic Wiki framework facilitates the collaboration in ontology co-authoring and sharing for people, and at the same time, makes it possible for the agent software to easily manage the ontology information. Eventually, the Semantic Wiki system accomplishes various tasks including the semantic view, the semantic navigation, and the semantic query.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.42 no.3 s.303
• /
• pp.9-16
• /
• 2005

For designing the database more efficiently, normailzation can be enforced to minimize the degree of unnecessary data redundancy and contribute to enhance data integrity. However, deep normalization tends to provoke multiple way of schema join, which could then induces response time degradation. To mitigate this sort of side effect that the normalization could brought, a number of field studies we observed adopted the idea of denormalization. To measure whether denormalization contributes to response time improvement, we in this paper developed two different data models about customer service system, one with perfect normalization and the other with denormalization, and evaluated their query response time behaviors. Performance results show that normalization case consistently outperforms denormalization case in terms of response time. This study show that the idea of denormalization, quite rarely contributes to that sort of improvement due ironically to the unnecessary data redundancy.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.33 no.1A
• /
• pp.67-73
• /
• 2008

Our research is aimed at developing an architectural frame-work of USN sensor network discovery service systems. The research is fo-cused on the four areas a survey of USN technology, development of a USN software model, development of the design space of the USN sensor network discovery service, and finally the architectural framework of the USN sensor network dicovery service. The survey of the USN technology is conducted on four technological visions that contain USN system technology, USN networking technology, and USN middleware along with the service platform, With respect to each technological division, domestic and worldwide leading research projects are primarily explored with their technical features and research projects are primarily explored with their technical features and research output To provide a means to analyze sensor network discovery services, we devel-oped the design space of the sensor network discovery services by exploring the scalability with respect to query scope, lookup performance, and resolution network.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2004.05b
• /
• pp.345-349
• /
• 2004

In order to index exactly moving objects(vehicle, mobile phone, PDA, etc.) in the mobile database, continuous updates of their locations are inevitable as well as time-consuming. The studies of pure spatial indices have focused on the efficient retrievals. However, the acquisition and management of the terminal Location of moving objects are more important than the efficiency of the query processing in the moving object databases. Therefore, it will be need to adopt parallel processing system for the moving object databases which should maintain the object's current location as precise as possible. This paper proposes a architecture of spatial indexing mobile objects using multiple processors. More precisely, we newly propose a method of splitting buckets using the properties of moving objects in order to minimize the number of database updates. We also propose a acquisition method for gathering the location information of moving objects and passing the information of the bucket extents in order to reduce the amount of passed messages between processors.

• Korean Journal of Community Nutrition
• /
• v.3 no.3
• /
• pp.410-422
• /
• 1998

To promote health status, strategies and interventions to improve nutrition should be based on the proper diagnosis of the subject's eating patterns. The elderly usually have traditional food habits and preferences, and it is very difficult to change them. This study was designed to identify dietary behavior and food preference of the elderly, in order to provide baseline data for the Elderly Nutrition Intervention Program for the Public Health Center. A survey questionnaire was made for use by trained interviewers to query 151elderly people from 5 community elderly centers located in Suwon, Korea. The majority of them ate regularly and partook of all available side dishes. Their major dietary problems were frequent consumptions of salty foods, and eating too quickly. They consumed grains and vegetables regularly, but seldomly ate dairy products, fruits, meat and food prepared with oil. They also tended to eschew ready made processed food, high cholesterol food, and fast food. Also they did not dine out as much as younger people. Desirable eating habit score were not significantly influenced by socioeconomic variables and nutrition-related characteristics. These included nutrition knowledge, Nutritional Risk Index(NRI) and a score of health concerns. However, meal balance scores were significantly higher in the younger group(p<.05), the higher household income group(p<.05). According to stepwise multiple regression analysis, NRI was the most important determinant of a desirable eating habit score for the male elderly, whereas the score of health concerns was mo9st important for female elderly subjects. The greatest predictor of the meal f balance score was nutrition knowledge. The elderly liked sweet tasting food, grains, rice, stews and Korean style soups. They disliked sour food, dairy products, processed food, and bread. The results indicate that the Elderly Nutrition Education Program should focus on increasing consumption of dairy products, fruits and food with oil, prepared by traditional Korean cooking methods. It also suggests that the program planning should consider the socioeconomic status of the elderly, such as income and education level, as well as concern for health.

• Journal of Broadcast Engineering
• /
• v.27 no.4
• /
• pp.581-592
• /
• 2022

Recently, as the requirements for using big data have increased, interest in dataset search technology needed for data analysis is also growing. Although it is necessary to proactively utilize metadata, unlike conventional text search, research on such dataset search systems has not been actively carried out. In this paper, we propose a new dataset-tailored search system that indexes metadata of datasets and performs dataset search based on metadata indices. The ranking given to the dataset search results from a newly devised algorithm that reflects the unique characteristics of the dataset. The system provides the capability to search for additional datasets which correlate with the dataset searched by the user-submitted query so that multiple datasets needed for analysis can be found at once.

• Journal of KIISE
• /
• v.44 no.10
• /
• pp.1112-1123
• /
• 2017

As content providers further offload content-centric services to the cloud, data retrieval over the cloud typically results in many redundant items because there is a prevalent near-duplication of content on the Internet. Simply fetching all data from the cloud severely degrades efficiency in terms of resource utilization and bandwidth, and data can be encrypted by multiple content providers under different keys to preserve privacy. Thus, locating near-duplicate data in a privacy-preserving way is highly dependent on the ability to deduplicate redundant search results and returns best matches without decrypting data. To this end, we propose an efficient near-duplicate detection scheme for encrypted data in the cloud. Our scheme has the following benefits. First, a single query is enough to locate near-duplicate data even if they are encrypted under different keys of multiple content providers. Second, storage, computation and communication costs are alleviated compared to existing schemes, while achieving the same level of search accuracy. Third, scalability is significantly improved as a result of a novel and efficient two-round detection to locate near-duplicate candidates over large quantities of data in the cloud. An experimental analysis with real-world data demonstrates the applicability of the proposed scheme to a practical cloud system. Last, the proposed scheme is an average of 70.6% faster than an existing scheme.