Juhyun Shin

A term normalization method for efficient knowledge acquisition through text processing

The importance of research on knowledge management is growing due to recent issues on Big Data. One of the most fundamental steps in knowledge management is the extraction of terminologies. Terms are often expressed in various forms and the variations often play a negative role, becoming an obstacle which causes knowledge systems to extract unnecessary ones. To solve the problem, we propose a method of term normalization which finds a normalized form (original and standard form defined in dictionaries) of variant terms. The method employs two characteristics of terms: appearance similarity measuring how similar terms are, context similarity measuring how many clue words they share. Through experiment, we show its positive influence of both similarities in term normalization.

Knowledge representation for video assisted by domain-specific ontology

Video analysis typically has been pursued in two different directions. Either previous approaches have focused on low-level descriptors, such as dominant color, or they have focused on the video content, such as person or object. In this paper, we present a video analysis environment not only to bridge these two directions but also can extract and manage semantic metadata from multimedia content autonomously for addressing the interaction between browsing and search capabilities. Concretely speaking, we implemented a tool that links MPEG-7 visual descriptors to high-level, domain-specific concepts. Our approach is ontology-driven, in the sense that we provide ontology based domain-specific extensions of the standards for describing the knowledge of video content. In this work, we consider one shot (episode) in the billiard game of video as the specific domain and we will be through the practical works to explain the process of representation of video knowledge. In the experiment part, we prove our approach effectiveness by comparing with the video content retrieval based on only key-word.

Adaptive-Tangent space representation for image retrieval based on kansei

From the engineering aspect, the research on Kansei information is a field aimed at processing and understanding how human intelligence processes subjective information or ambiguous sensibility and how such information can be executed by a computer. Our study presents a method of image processing aimed at accurate image retrieval based on human Kansei. We created the Kansei-Vocabulary Scale by associating Kansei of high-level information with shapes among low-level features of an image and constructed the object retrieval system using Kansei-Vocabulary Scale. In the experimental process, we put forward an adaptive method of measuring similarity that is appropriate for Kansei-based image retrieval. We call it “adaptive-Tangent Space Representation (adaptive-TSR)”. The method is based on the improvement of the TSR in 2-dimensional space for Kansei-based retrieval. We then it define an adaptive similarity algorithm and apply to the Kansei-based image retrieval. As a result, we could get more promising results than the existing method in terms of human Kansei.

The Method of Semi-supervised Automatic Keyword Extraction for Web Documents using Transition Probability Distribution Generator

In this paper, a semi-supervised method for automatic keyword extraction of web documents using unconventional Markov Chain with conditional transition matrices for each distinct feature distributed by Transition Probability Distribution Generator (TPDG) is introduced. Since keywords are the set of the most appropriate and relevant words which define the context of the document precisely and concisely, many applications such as text data mining, text analytics and other natural language processes of deriving high-quality information from text can take advantage of it. The conditional transition matrices for each distinct feature of the model is the state-of-the-art which mostly rely on the characteristics of the keywords and distribution probabilities of each feature on the state space in order to learn the sequence of behaviors of the keywords in various web documents. According to the experimental results, the proposed method outperforms the baseline methods for keyword extraction in terms of performance and semantically.

A Method for Recommending the Most Appropriate Expansion of Acronyms Using Wikipedia

Over the years, many researchers have been studied to detect expansions of acronyms in texts by using linguistic and syntactical approaches in order to overcome disambiguation problems. Acronym is an abbreviation formed which is composed of initial components of single or multiple words. These initial components bring huge mistakes when a machine conducts experiments to find meaning from given texts. Detecting expansions of acronyms is not a big issue now days. The problem is that a polysemous acronym. In order to solve this problem, this paper proposes a method to recommend the most related expansion of acronym through analyzing co-occurrence words by using Wikipedia. Our goal is not finding acronym definition or expansion but recommending the most appropriate expansion of given acronyms.

A term normalization method for better performance of terminology construction

The importance of research on knowledge management is growing due to recent issues with big data. The most fundamental steps in knowledge management are the extraction and construction of terminologies. Terms are often expressed in various forms and the term variations play a negative role, becoming an obstacle which causes knowledge systems to extract unnecessary knowledge. To solve the problem, we propose a method of term normalization which finds a normalized form (original and standard form defined in dictionaries) of variant terms. The method employs a couple of characteristics of terms: one is appearance similarity, which measures how similar terms are, and the other is context similarity which measures how many clue words they share. Through experiment, we show its positive influence of both similarities in the term normalization.

Semantic Representation of Motion for Tracing Object in Surveillance System

Recent proliferation of multimedia data necessitates effective and efficient methods of retrieving of multimedia data. Especially, intelligent surveillance system needs semantic multimedia data processing. However, these studies are significantly insufficient to enable a complete semantic recognition in regard to semantic-based video retrieval. Current technology focuses on analysis of low-level characteristics such as color, texture, shape, and trajectory of an object and on applications of those characteristics. In this paper, we develop a semantic representation in human language to reduce the semantic gap between low-level and high-level characteristics; with consideration of not only the low-level characteristics but also the high-level characteristics with the use of the human language. Particularly, this paper focuses on semantic representation by using topological and directional relations between non-moving and moving objects for intrusion detection system using Closed-Circuit Television (CCTV). To this end, we capitalize on size relation of the object based on the previous studies to suggest 3D spatio-temporal relation. We also intend to reduce semantic gap by mapping with the use of predicates. This paper extends our previous work [17] published on SeCIHD 2011.

Extended spatio-temporal relations between moving and non-moving objects

During the last decade, the emerging technology for video retrieval is mainly based on the content. However, semantic-based video retrieval has become more and more necessary for the humans especially the naive users who can only use the human language during retrieval. In this paper, we focus on semantic representation using topological and directional relations between non-moving and moving objects for security using CCTV(closed-circuit Television). In this paper, we propose new spatio-temporal relation to extend previous work using topological and directional relations and investigate spatiotemporal predicates which propose our models. In the experiment part, we compared retrieval results using TSR(Tangent Space Representation) with those using rules represented by the proposed model.

An event-description framework for video application using spatio-temporal relations

Video applications usually involve a large number of moving objects. Moving objects refer to semantic real-world entity definitions that are used to denote a coherent spatial region and be automatically computed by the continuity of spatial low-level features, such as color and motion. Spatial and temporal relationships among these objects should be efficiently supported and retrieved within a video authoring tool. In this paper we present several spatial, temporal and spatio-temporal relationships of interest and propose efficient indexing scheme, based on multidimensional spatial data structures, for video applications that involve objects. So, we emphasize on analyzing and interpreting video object motions for advanced video application. To realize this objective the research in this field is subdivided into two main directions: (1) Moving objects description at the low levels: using the spatio-temporal relationships to analyze and present the video object motions. (2) Moving object description at the semantic level: Actions, events and interactions of moving objects.