• Asian Journal of Innovation and Policy
• /
• v.7 no.1
• /
• pp.150-177
• /
• 2018

We review two specific risk management strategies of venture capitalists (VCs): deal syndication and domain specialization with respect to their explicit role in adjudging and managing the overall magnitude of information asymmetry risks. These are analyzed for three distinct categories of VC firms as classified by their funding stage focus (early vs. late), ownership type (foreign vs. domestic) and the human capital composition of the core VC team (entrepreneurial vs. investor). The analysis is based on both secondary data and primary data for active 72 VC firms in India. Syndication is moderately important for entrepreneurial VC firms, but not at all important for early-stage focused and foreign VC firms. This finding is distinctly different from what has been conventionally observed in the literature. Among the various arenas of domain specialization, high-technology focus is important for all segments of VC firms. In the context of investment-stage focus, foreign VC firms exhibit growth-stage specialization, while entrepreneurial VC firms concentrate on earlier investment stages.

• Korean Journal of Cognitive Science
• /
• v.10 no.1
• /
• pp.67-78
• /
• 1999

Information providers on WWW have been rapidly increasing, and they provide a vast amount of information in various fields, Because of this reason, it becomes hard for users to get the information they want. Although there are several search engines that help users with the keyword matching methods, it is not easy to find suitable keywords. In order to solve these problems with a specific domain, we propose an intelligent information agent(HHA-la : HongIk Information Agent) that converts user's q queries to forms including related domain words in order to represent user's intention as much as it can and provides the necessary information of the domain to users. HHA-la h has an ontological knowledge base of animal domain, supplies necessary information for queries from users and other agents, and provides relevant web page information. One of system components is a WebDB which indexes web pages relevant to the animal domain. The system also supplies new operators by which users can represent their thought more clearly, and has a learning mechanism using accumulated results and user feedback to behave more intelligently, We implement the system and show the effectiveness of the information agent by presenting experiment results in this paper.

• ETRI Journal
• /
• v.29 no.2
• /
• pp.179-188
• /
• 2007

We propose a multi-strategic concept-spotting approach for robust spoken language understanding of conversational Korean in a hostile recognition environment such as in-car navigation and telebanking services. Our concept-spotting method adopts a partial semantic understanding strategy within a given specific domain since the method tries to directly extract predefined meaning representation slot values from spoken language inputs. In spite of partial understanding, we can efficiently acquire the necessary information to compose interesting applications because the meaning representation slots are properly designed for specific domain-oriented understanding tasks. We also propose a multi-strategic method based on this concept-spotting approach such as a voting method. We present experiments conducted to verify the feasibility of these methods using a variety of spoken Korean data.

• Journal of the Korean Home Economics Association
• /
• v.48 no.1
• /
• pp.41-54
• /
• 2010

The purpose of this study was to investigate the relationships among children's important self-domain, self-evaluation and global self-esteem by gender. Information was collected on 376 5-6th graders from elementary schools in Busan. The major findings were that 1) The self-evaluation of peer domain was the most powerful determinant on self-esteem in both boys and girls 2) The percentage that valued domain of family self was higher than the other groups in both boys and girls(boys 36.3%, girls 55.4%). 3) In boys' domains of peer self and computer self, important self-domain moderated the effect of self evaluation on global self-esteem.

• Journal of KIISE:Databases
• /
• v.27 no.1
• /
• pp.64-78
• /
• 2000

In this paper, we design and implement a system for managing domain specific thesauri, where object-oriented paradigm is applied to thesaurus construction, concept browsing and query-based reference. This system provides an objected-oriented mechanism to assist domain experts in constructing thesauri; it determines a considerable part of relationship degrees between terms by inheritance and supplies domain experts with information available from a thesaurus being constructed This information is especially useful to enforce consistency between the hierarchies of a thesaurus, each constructed by different experts in different sites through cooperation. It may minimize the burden of domain eIn this paper, we design and implement a system for managing domain specific thesauri, where object oriented paradigm is applied to thesaurus construction, concept browsing and query based reference. This system provides an objected mechanism to assist domain experts in constructing thesauri: it determines a considerable part of relationship degrees between terms by inheritance and supplies domain experts with information available from a thesaurus being constructed. This information is especially useful to enforce consistency between the hierarchies of a thesaurus, each constructed by different experts in different sites through cooperation. It may minimize the burden of domain experts caused from the exhaustive specification of individual relationship. This system also provides an abstracted browsing and a query based reference, which allow users to easily verify thesaurus terms before they are used in usual boolean queries. The verification is made by actively searching for them in the thesaurus. Reference queries and abstracted browsing views facilitate this searching. The facility is indispensable especially when precision counts for much.

• Journal of Korea Society of Industrial Information Systems
• /
• v.29 no.3
• /
• pp.79-87
• /
• 2024

Recently, the importance of pre-trained language models (PLM) has been emphasized for natural language processing (NLP) such as text classification, sentiment analysis, and question answering. Korean PLM shows high performance in NLP in general-purpose domains, but is weak in domains such as finance, medicine, and law. The main goal of this study is to propose a language model learning process and method to build a financial-specific language model that shows good performance not only in the financial domain but also in general-purpose domains. The five steps of the financial-specific language model are (1) financial data collection and preprocessing, (2) selection of model architecture such as PLM or foundation model, (3) domain data learning and instruction tuning, (4) model verification and evaluation, and (5) model deployment and utilization. Through this, a method for constructing pre-learning data that takes advantage of the characteristics of the financial domain and an efficient LLM training method, adaptive learning and instruction tuning techniques, were presented.

• Journal of Information Processing Systems
• /
• v.17 no.5
• /
• pp.867-878
• /
• 2021

In context awareness and user intention tasks, dataset construction is expensive because specific domain data are required. Although pretraining with a large corpus can effectively resolve the issue of lack of data, it ignores domain knowledge. Herein, we concentrate on data domain knowledge while addressing data scarcity and accordingly propose a multi-channel long short-term memory (LSTM). Because multi-channel LSTM integrates pretrained vectors such as task and general knowledge, it effectively prevents catastrophic forgetting between vectors of task and general knowledge to represent the context as a set of features. To evaluate the proposed model with reference to the baseline model, which is a single-channel LSTM, we performed two tasks: voice phishing with context awareness and movie review sentiment classification. The results verified that multi-channel LSTM outperforms single-channel LSTM in both tasks. We further experimented on different multi-channel LSTMs depending on the domain and data size of general knowledge in the model and confirmed that the effect of multi-channel LSTM integrating the two types of knowledge from downstream task data and raw data to overcome the lack of data.

• Journal of KIISE:Software and Applications
• /
• v.37 no.4
• /
• pp.297-304
• /
• 2010

Clustering method which based on sentence type or document genre is a technique used to improve translation quality of SMT(statistical machine translation) by domain-specific translation. But there is no previous research using sentence type and document genre information simultaneously. In this paper, we suggest an integrated clustering method that classifying sentence type by syntactic structure similarity and document genre by word similarity information. We interpolated domain-specific models from clusters with general models to improve translation quality of SMT system. Kernel function and cosine measures are applied to calculate structural similarity and word similarity. With these similarities, we used machine learning algorithms similar to K-means to clustering. In Japanese-English patent translation corpus, we got 2.5% point relative improvements of translation quality at optimal case.

• Proceedings of the Korea Database Society Conference
• /
• 1999.06a
• /
• pp.281-287
• /
• 1999

Decision class analysis (DCA) is viewed as a classification problem where a set of input data (situation-specific knowledge) and output data (a topological leveled influence diagram (ID)) is given. Situation-specific knowledge is usually given from a decision maker (DM) with the help of domain expert(s). But it is not easy for the DM to know the situation-specific knowledge of decision problem exactly. This paper presents a methodology fur sensitivity analysis of DCA under incomplete information. The purpose of sensitivity analysis in DCA is to identify the effects of incomplete situation-specific frames whose uncertainty affects the importance of each variable in the resulting model. For such a purpose, our suggested methodology consists of two procedures: generative procedure and adaptive procedure. An interactive procedure is also suggested based the sensitivity analysis to build a well-formed ID. These procedures are formally explained and illustrated with a raw material purchasing problem.

• Journal of the Korean Society for Library and Information Science
• /
• v.54 no.3
• /
• pp.263-283
• /
• 2020

This study proposed a method to redesign the search system by extracting the class, property, and code values of value domain from the key elements of the MDR system that conforms to the ISO/IEC 11179 standard. To implement this, an operating system conforming to the standard is required. Therefore, this study targeted the MDR system of the Korea Information and Communication Technology Association. As a result, it was confirmed that the redesigned MDR system can easily search and utilize specific metadata without understanding the conceptual domain, metadata element concept, metadata element, and value domain proposed in the standard.