Classification Technique for Filtering Sentiment Vocabularies for the Enhancement of Accuracy of Opinion Mining

Ji-Hoon Seo,Ho-Sun Lee,Jin-Tak Choi 보안공학연구지원센터 2015 International Journal of u- and e- Service, Scienc Vol.8 No.10

This thesis, as part of the creation of a text-mining-based sentiment dictionary to be applied in the Korean grammar structure, solves the problem of the enhancement of accuracy of opinion mining data by applying the filtering model of candidate sentiment vocabularies. The fact that the reliability of sensitive vocabularies shows huge variances according to the filtering modeling method applied has become a decreasing factor for the accuracy of the vocabularies in the opinion mining process, which is attributable to the fact there isn’t a success factor in the filtering modeling standard for precise selection of vocabularies. In this thesis, a filtering model of positive and negative vocabularies on candidate Korean sentiment vocabularies and a reliability scale for accuracy were suggested to solve such problems by applying the semi-structured data filtering model for the selection of candidate sentiment vocabularies of the Korean grammar. The study has confirmed through relevant performance assessment when filtering were applied in relation to 30%, 50% and 60% respectively with regard to candidate sentiment vocabularies upon collecting vocabularies obtained via sentence segmentation and classification into positive and negative vocabularies that exceptional accuracy of the opinion sentiment dictionary was shown via the 60% filtering.

A Temporal Microblog Filtering Model

Zhongyuan Han,Muyun Yang,Leilei Kong,Haoliang Qi,Sheng Li 보안공학연구지원센터 2016 International Journal of Grid and Distributed Comp Vol.9 No.1

The rapid growth in the popularity of social networking and microblogging has led to a new way of finding and broadcasting information in the recent years. The real-time microblog filtering emerges as the times require. The task of real-time microblog filtering is to decide if subsequently posted tweets are relevant to a given query which represents the special information needs. One-side feedback is one of the most difficult problems in microblog filtering. This paper focuses on exploiting the time profile of relevant microblogs to address this problem. A temporal microblog filtering based on retrieval model is proposed. Specifically, similarity threshold achieved by the language model is adjusted according to temporal burst. Evaluated on the TREC 2012 microblog real-time filtering track dataset, the experimental results show that the proposed model is significantly better than several baselines.

데이터 공학 : 가중 윈도우를 통한 사용자 이력 기반 추천 시스템

황성민 ( Sungmin Hwang ),( Rajashree Sokasane ),( Hiep Tuan Nguyen Tri ),김경백 ( Kyungbaek Kim ) 한국정보처리학회 2015 정보처리학회논문지. 소프트웨어 및 데이터 공학 Vol.4 No.6

온라인에서 물품을 구매하고자 할 때, 추천 시스템은 사용자에 맞춘 추천을 하게 되며, 사용자가 관심을 가질만한 새로운 물품까지 추천해준다. Collaborative filtering 등, 여러 모델들이 보다 정확한 추천을 위해 제안되었으며, 활발히 연구되고 있다. 그중 Collaborative filtering은 사용자 선호도를 예측하는 데 좋은 결과를 보여주지만 사용자 개체 및 데이터가 부족한 환경에서는 사용자들끼리의 비교를 힘들게하여 collaborative filtering이 적용되기 힘들게 한다. 새로 시작하는 시스템이거나 사용자 개체 수가 적은 경우, 문제가 발생하며, 이와 같은 상황에서는 content-based filtering이 사용된다. 하지만 content-based filtering은 비슷한 물건만 추천해주거나, 사용자 성향 변화를 제대로 반영하지 못하는 등의 여러 단점을 가지고 있다. 이러한 한계점들을 극복하기 위해서 사용자 구매 기록에 가중 윈도우를 적용하고, 사용자 구매 기록 분석을 통한 윈도우 가중치 조정을 수행하는 시스템을 제안한다. 사용자 성향의 변화에 민감하게 반응할 수 있고, 이를 기반으로 무의미한 추천을 제거하며, 사용자가 찾기 어려운 관련 물품 추천이 가능한 새로운 상품도 추천하는 시스템을 제시하며, 언급된 사용자 개체 및 데이터가 부족한 환경에서의 동작을 검증하기 위해, 스타트업 무역업체에서 제공된 상품정보 기반 실험을 통해 제안된 시스템의 동작을 검증하였다. When we buy items in online stores, it is common to face recommended items that meet our interest. These recommendation system help users not only to find out related items, but also find new things that may interest users. Recommendation system has been widely studied and various models has been suggested such as, collaborative filtering and content-based filtering. Though collaborative filtering shows good performance for predicting users preference, there are some conditions where collaborative filtering cannot be applied. Sparsity in user data causes problems in comparing users. Systems which are newly starting or companies having small number of users are also hard to apply collaborative filtering. Content-based filtering should be used to support this conditions, but content-based filtering has some drawbacks and weakness which are tendency of recommending similar items, and keeping history of a user makes recommendation simple and not able to follow up users preference changes. To overcome this drawbacks and limitations, we suggest weighted window assisted user history based recommendation system, which captures user’s purchase patterns and applies them to window weight adjustment. The system is capable of following current preference of a user, removing useless recommendation and suggesting items which cannot be simply found by users. To examine the performance under user and data sparsity environment, we applied data from start-up trading company. Through the experiments, we evaluate the operation of the proposed recommendation system.

고속 RFID 필터링 엔진의 설계와 캐쉬 기반 성능 향상

박현성,김종덕,Park Hyun-Sung,Kim Jong-Deok 한국통신학회 2006 韓國通信學會論文誌 Vol.31 No.5a

본 논문은 다수의 RFID 태그가 사용되고 있는 환경에서 고속 필터링을 수행하기 위한 필터링 엔진을 설계한다. 이를 위하여 우리는 고속 라우터나 방화벽에 적용되었던 고속 패킷 필터링 기법이 RFID 데이터 필터링과 매우 유사함을 보이고 그 중 대표적인 기법인 Bit Parallelism 기반의 Aggregated Bit Vector(ABV)를 고속 RFID 필터링 엔진에 적용한다. 또한, RFID 데이터 필터링의 성향을 관찰한 결과 태그 인식 및 필터 부합의 시간적 중복성을 발견하고 두 가지 캐쉬(태그 캐쉬, 필터 캐쉬)를 적용하여 추가적인 필터링 성능 향상을 꾀하였다. 설계한 RFID 고속 필터링 엔진의 성능 평가를 위해 프로토타입 애플리케이션을 제작하여 시뮬레이션을 수행하였다. 결과로써 기존의 순차적인 RFID 데이터 필터링에 비해 고속의 필터링 성능을 보이며 특히 필터의 수가 증가할수록 필터링의 효율이 높아짐을 보인다. In this paper, we present a high-speed RFID data filtering engine designed to carry out filtering under the conditions of massive data and massive filters. We discovered that the high-speed RFID data filtering technique is very similar to the high-speed packet classification technique which is used in high-speed routers and firewall systems. Actually, our filtering engine is designed based on existing packet classification algorithms, Bit Parallelism and Aggregated Bit Vector(ABV). In addition, we also discovered that there are strong temporal relations and redundancy in the RFID data filtering operations. We incorporated two kinds of caches, tag and filter caches, to make use of this characteristic to improve the efficiency of the filtering engine. The performance of the proposed engine has been examined by implementing a prototype system and testing it. Compared to the basic sequential filter comparison approach, our engine shows much better performance, and it gets better as the number of filters increases.

Two-Stage Frequency-Temporal Filtering for Effective Audio Fingerprinting of Sound Recordings

Mansoo Park,김회린,Seung Hyun Yang 에스케이텔레콤 (주) 2009 Telecommunications Review Vol.19 No.3

Sound recordings are commonly distorted by channel and background noise. The performance of audio identification is mainly degraded by that noise. For an audio fingerprinting system, Haitsma and Kalker introduced a robust and efficient audio hashing scheme applying high-pass filtering (differentiation) to the frequency-temporal sequence of perceptual filter-bank energies. However, the robustness of the audio fingerprinting scheme is still important in real noisy environments. This paper introduces some alternatives of frequency-temporal filtering for effective audio fingerprinting of sound recordings in real environments. As the alternative to frequency filtering, a type of band-pass filter, instead of a high-pass filter, is used to enhance robustness to background noise in a real situation. As the alternative to temporal filtering, RASTA, instead of a high-pass filter, is used for normalizing sound recording conditions. As well, this paper introduces a two-stage audio fingerprinting scheme to achieve synergy of the combination of frequency-temporal filtering. Experimental results show that the proposed method is effective for sound recordings in real environments.

복합 필터링을 이용한 IPTV-VOD 프로그램 추천 시스템 연구

강용진,선철용,박규식 대한전자공학회 2010 電子工學會論文誌-SP (Signal processing) Vol.47 No.4

본 연구는 IPTV 환경에서 사용자의 취향에 맞는 VOD 프로그램을 추천할 수 있는 시스템을 새로이 제안하였다. 제안 시스템은 내용기반 필터링과 협업 필터링의 장․단점을 상호 보완한 복합 필터링에 의한 IPTV-VOD 프로그램 추천 시스템으로, 각 필터링 기법의 프로그램 선호도(program preference) 값을 단일 지표(single-scale)로 비교․평가할 수 있는 수단을 제공함으로써 실질적인 복합 필터링 추천 시스템을 구축하였다. 사용자의 프로그램 선호 취향을 나타내는 사용자 프로파일(user profile)은 사용자의 과거 프로그램 시청 이력뿐만 아니라 사용자와 유사한 이웃 사용자들의 취향을 1주일 단위로 갱신되는 프로그램 선호도와 중분류 선호도로 표현하였기 때문에 보다 정확한 프로그램 추천이 가능하다. 제안 시스템의 성능평가를 위해 시청률 조사기관인 닐슨리서치의 24주분 지상파 및 케이블 방송 시청 데이터를 IPTV 형식에 맞게 재구성하여 사용하였으며, 다양한 실험을 통해 그 실용성을 입증하였다. In this paper, a new program recommendation system is proposed to recommend user preferred VOD program in IPTV environment. A proposed system is implemented with hybrid filtering method that can cooperatively complements the shortcomings of the content-based filtering and collaborative filtering. For a user program preference, a single-scaled measure is designed so that the recommendation performance between content-based filtering and collaborative filtering is easily compared and reflected to final hybrid filtering procedure. In order to provide more accurate program recommendation, we use not only the user watching history, but also the user program preference and sub-genre program preference updated every week as a user preference profile. System performance is evaluated with modified IPTV data from real 24-weeks cable TV watching data provided by Nilson Research Corp. and it shows quite comparative quality of recommendation.

디지털 영상의 픽셀값 경사도에 의한 미디언 필터링 포렌식 판정

이강현(Kang Hyeon RHEE) 대한전자공학회 2015 전자공학회논문지 Vol.52 No.6

디지털 영상의 배포에서, 위·변조자에 의해 영상이 변조되는 심각한 문제가 있다. 이러한 문제를 해결하기 위하여, 본 논문에서는 영상의 픽셀값 경사도에 따른 특징벡터를 이용한 미디언 필터링 영상 포렌식 판정 알고리즘을 제안한다. 제안된 알고리즘에서, 원영상의 픽셀값 경사도로부터 자기회귀 계수를 1∼6차까지의 6 Dim.을 계산한다. 그리고 경사도를 Poisson 방정식의 해에 의한 재구성 영상과 원영상과의 차영상으로 부터, 4 Dim. (평균값, 최대값 그리고 최대값의 좌표 i,j)의 특징벡터를 추출한다. 2 종류의 특징벡터는 10 Dim.으로 조합되어 변조된 영상의 미디언 필터링 (Median Filtering: MF) 검출기의 SVM(Support Vector Machine) 분류를 위한 학습에 사용된다. 제안된 미디언 필터링 검출 알고리즘은 동일 10 Dim. 특징벡터의 MFR (Median Filter Residual) 스킴과 비교하여 원영상, 평균필터링 (3×3) 영상 그리고 JPEG (QF=90) 영상에서는 성능이 우수하며, Gaussian 필터링 (3×3) 영상에서는 성능이 다소 낮지만, 성능평가 전체항목에서 민감도 (Sensitivity; TP: True Positive rate)와 1-특이도 (1-Specificity; FP: False Positive rate)의 AUC (Area Under Curve)가 모두 1에 수렴하여 ‘Excellent (A)’ 등급임을 확인하였다. In a distribution of digital image, there is a serious problem that is a distribution of the altered image by a forger. For the problem solution, this paper proposes a median filtering (MF) image forensic decision algorithm using a feature vector according to the pixel value"s gradients. In the proposed algorithm, AR (Autoregressive) coefficients are computed from pixel value" gradients of original image then 1th∼6th order coefficients to be six feature vector. And the reconstructed image is produced by the solution of Poisson"s equation with the gradients. From the difference image between original and its reconstructed image, four feature vector (Average value, Max. value and the coordinate i,j of Max. value) is extracted. Subsequently, Two kinds of the feature vector combined to 10 Dim. feature vector that is used in the learning of a SVM (Support Vector Machine) classification for MF (Median Filtering) detector of the altered image. On the proposed algorithm of the median filtering detection, compare to MFR (Median Filter Residual) scheme that had the same 10 Dim. feature vectors, the performance is excellent at Unaltered, Averaging filtering (3×3) and JPEG (QF=90) images, and less at Gaussian filtering (3×3) image. However, in the measured performances of all items, AUC (Area Under Curve) by the sensitivity and 1-specificity is approached to 1. Thus, it is confirmed that the grade evaluation of the proposed algorithm is "Excellent (A)".

Research on Image Restoration Algorithm for Dynamic Target

Hua Xiang 보안공학연구지원센터 2015 International Journal of Signal Processing, Image Vol.8 No.8

For movement target, image recognition with blur is a difficult problem to solve in image acquisition system. The paper proposes an image restoration algorithm based on inverse filtering and wiener filtering. According to original image of movement target by acquisition system, it analyses frequency spectrum characteristic, builds restoration model for image degradation, researches inverse filtering and wiener filtering technologies of image preprocessing algorithm and image restoration algorithm. Further it explains image quality assessment method, discusses objective assessment method without reference in detail. Dealing with the movement target image of actual acquisition, some images are processed based on the restoration method which proved. By contrasting algorithm with result, the restoration method is valid to eliminate blur and noise of disturbing influence from target moving, which proposed combining with inverse filtering and wiener filtering. The result shows that the edge characteristic is obvious and propitious to reflect original target actually.

순환신경망 기법을 이용한 스파 플랫폼의 시계열데이터 필터링에 관한 연구

유승열,이재철,이종현,황호진,이순섭 한국마린엔지니어링학회 2019 한국마린엔지니어링학회지 Vol.43 No.1

There is growing interest in the numerous techniques focused on analyzing vast quantities of measurement data in real time for the development of smart ships along with the development of asset integrity management systems for offshore platforms. To analyze the measurement data in real time, data filtering is used to eliminate the noise in the data and then extract the necessary information to perform a comprehensive data analysis. In the traditional shipbuilding and offshore industry, spectrum-based filtering methods are used because the corresponding data is saved for a certain period and subsequently analyzed. These methods are not suited to the present situation in which real-time data is required to be analyzed. Therefore, a new method for data filtering is required. The objective of this study is to filter data in real time using a recurrent neural network algorithm, which is a deep learning model used for learning time series data. In order to filter the measured mooring tension value of the spa platform in real time, a filtering model comprising a recurrent neural network algorithm was designed, and the results of the data filtering process were verified to confirm the possibility of real- time filtering. 스마트 선박 (Smart ship)의 개발과 해양 플랫폼의 예지보전 시스템 및 자산 관리 시스템 개발을 위해 방대한 양의계측 데이터를 실시간으로 분석할 수 있는 기술에 대한 관심이 높아지고 있다. 이러한 계측 데이터를 실시간으로 분석하기 위해서는 계측 데이터의 노이즈를 제거하고 필요한 정보를 추출하여 분석에 용이한 형태로 데이터를 가공하는 과정인 데이터 필터링이 반드시 선행되어야 한다. 기존의 조선 해양 산업에서는 일정기간 이상 데이터를 저장한 후 이에대한 분석을 실시하여 스펙트럼 기반의 필터링 기법을 많이 이용하였다. 이러한 방법은 실시간 데이터를 분석해야 하는현 상황에는 적합하지 않아 실시간 데이터를 필터링하기 위한 새로운 기법이 필요한 실정이다. 본 논문에서는 시계열데이터를 학습하기 위한 딥 러닝 모델인 순환신경망 알고리즘을 이용하여 실시간으로 전송되는 데이터를 필터링하고자하였다. 실시간으로 계측되는 스파 플랫폼의 계류 장력 값을 필터링하기 위해 순환신경망 알고리즘을 이용한 필터링 모델을 설계하고 그 결과값을 확인하여 실시간 필터링 가능 여부를 확인하였다. 최종적으로 실시간으로 전송되는 데이터를 필터링 하기 위해 순환신경망 알고리즘을 사용하는 것이 적합하다는 것을 확인하였다.

윤활유 필터의 종류 및 특징

홍성호,신주용,박태성,이상후 한국트라이볼로지학회 2023 한국트라이볼로지학회지 (Tribol. Lubr.) Vol.39 No.4

This paper presents a discussion on lubricating oil filters. The maintenance of lubricating oil filters can improve the performance of mechanical systems and extend the service life of the lubricating oil. Therefore, the effective management of the lubricating oil can extend the service life of the machine and reduce maintenance costs. A representative method for managing lubricating oil is filtering the lubricating oil using a lubricant filter. However, effectively managing a lubricating oil using a lubricant filter requires an understanding of the related knowledge. In this paper, we present the definition, classification, characteristics, specifications, performance, and self-cleaning function of lubricating oil filters. The lubricant filters are classified based on the filter material, filtering method, filtering location, and amount of filtered fluid. Cellulose and glass fiber materials are conventionally used as materials for lubricant filters, and recently, metal materials, which show excellent durability, are being increasingly adopted. The filtering methods can be classified into physical, chemical, magnetic, and electric field methods, and the lubricant filters can be classified according to their location in the lubrication system. The beta ratio and efficiency of the lubricant filter can be determined based on the performance of the filter. Finally, there are many products or technologies that add a self-cleaning function to the filter to remove foreign substances or contaminants for efficient management.