• 한국컴퓨터정보학회논문지
• /
• 제27권11호
• /
• pp.57-66
• /
• 2022

본 논문에서는 Sentence-BERT 모델을 활용한 법률 문서 검색 방법을 제안한다. 법률 검색 서비스를 이용하고자 하는 일반인들은 법률 용어 및 구조에 대한 이해가 부족함에 따라 관련 판례 검색 등에 있어 어려움을 겪고 있다. 기존의 키워드 및 텍스트마이닝 기반 법률 검색 방법은 판결문의 문맥에 대한 정보가 없으며, 동음이의어 및 다의어에 대해 구분하기 어려워 성능을 높이는 데 한계가 있었다. 그로 인해 법률 문서 검색 결과에 대한 정확도가 낮아 신뢰하기가 어려웠다. 이를 위해, 대법원 판례 및 법률구조공단 상담사례 데이터에서 일반인의 법률 검색 문장에 대한 성능을 개선하고자 한다. Sentence-BERT 모델은 판례 및 상담 데이터에 대한 문맥 정보가 임베딩 되므로, 문장의 의미 손실이 적어 TF-IDF 및 Doc2Vec 검색 방법과 비교했을 때보다 검색 정확도가 개선된 것을 확인할 수 있었다.

• 대한음성학회지:말소리
• /
• 제50호
• /
• pp.123-137
• /
• 2004

In this paper, we consider the implementation of connected digit recognition system using tree structured lexicon model. To implement efficiently the fixed or variable length digit recognition system, finite state network (FSN) is required. We merge the word network algorithm that implements the FSN with lexical tree search algorithm that is used for general speech recognition system for fast search and large vocabulary systems. To find the efficient modeling of digit recognition system, we investigate some performance changes when the lexical tree search is applied.

• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• 제15권4호
• /
• pp.267-275
• /
• 2011

This paper proposes a new quickly convergent pattern search quasi-Newton algorithm that employs the multi-step version of the Symmetric Rank One (SRI). The new algorithm works on the factorizations of the inverse Hessian approximations to make available a sequence of convergent positive bases required by the pattern search process. The algorithm, in principle, resembles that developed in [1] with multi-step methods dominating the dervation and with numerical improvements incurred, as shown by the numerical results presented herein.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2000년도 하계종합학술대회 논문집(4)
• /
• pp.39-42
• /
• 2000

Motion estimation plays an important role for video coding. In this paper, we derive optimal search patterns for fast block matching motion estimation. By analyzing the block matching algorithm as a function of block shape and size, we can find an optimal search pattern for initial motion estimation. The proposed idea, which has been verified experimentally by computer simulations, can provide an analytical basis for the current MPEG-2 proposals. In order to choose a more compact search pattern for BMA, we exploit the statistical relationship between the motion and the frame difference of each block.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2000년도 추계종합학술대회 논문집(4)
• /
• pp.183-186
• /
• 2000

Motion estimation technique has been used to increase video compression rates in motion video applications. One of the important algorithms to implement the motion estimation technique is search algorithm. Among many search algorithms, the H.263 adopted the Nearest Neighbors algorithm for fast search. In this paper, motion estimation block for the Nearest Neighbors algorithm is designed on FPGA and coded using VHDL and simulated under the Xilinx foundation environments. In the experiment results, we verified that the algorithm was properly designed and performed on the Xilinx FPGA(XCV300Q240)

• 전력전자학회:학술대회논문집
• /
• 전력전자학회 2004년도 전력전자학술대회 논문집(2)
• /
• pp.926-930
• /
• 2004

This paper presents a comparison of the characteristics by two types of switched reluctance motor (SRM) drives using search coils, Exterior pole search coil and Interior central pole search coil, which are used for estimation of the rotor position. The advantages and disadvantages of the two types of drive methods are also described.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2003년도 하계종합학술대회 논문집 Ⅳ
• /
• pp.2044-2047
• /
• 2003

In this paper, we propose efficient motion vector refinement algorithm for frame-rate reduction transcoding. The proposed algorithm is to set the search range for motion refinement based on the incoming motion vector. The algorithm calculates the importance of motion vector of the skipped frame and then selects two motion vector to set search range. Through this process, we determine the accuracy of incoming motion vector and set the search range lot refinement adaptively by means of the accuracy. In experiments, we show efficiency of our algorithm to reduce the search points for refinement.

• 한국경영과학회지
• /
• 제10권2호
• /
• pp.75-82
• /
• 1985

Reorganization is essential part of data base maintenanc work and the reasonable reorganization points can be determined from the trade-off between reorganization cost and performance degradation. There has been many reorganization models so far, but none of these models have assumed nonlinear search cost function. This paper presents the existensions of two existing linear reorganization models for the case where the search cost function is quadratic. The higher performance of these extended models was shown in quadratic search cost function case.

• 한국경영과학회지
• /
• 제19권1호
• /
• pp.201-206
• /
• 1994

The linear search problem is concerned with finding a hiden target on the real line R. The position of the target governed by some probability distribution. It is desired to find the target in the least expected search time. This problem has been formulated as an optimization problem by a number of authors without making use of Markov Decision Process (MDP) theory. It is the aim of the paper to give a (MDP) formulation to the search problem which we feel is both natural and easy to follow.

• 한국통신학회논문지
• /
• 제21권2호
• /
• pp.313-325
• /
• 1996

This paper presents a new architecture for full-search block-matching motion estimation. The architecture is based on linear systolic arrays. High speed operation is obtained by feeding reference data, search data, and control signals into the linear systolic array in a pipelined fashion. Input data are fed into the linear systolic array at a half of the processor speed, reducing the required data bandwidth to half. The proposed architecture has a good scalability with respect to the number of processors and input bandwidth when the size of reference block and search range change.