Higher Recall Ratio Research Articles

Information retrieval algorithm of industrial cluster based on vector space

AbstractThe current information retrieval research on industrial clusters has low precision, low recall ratio, obvious delay and high energy consumption. Thus, in this paper, a information retrieval algorithm based on vector space for industrial clusters is proposed. By optimizing the unlawful labels in the database network, dividing the web pages of the industrial cluster information database and calculating the keyword scores of the relevant information of the industrial cluster corresponding to a web page, a set of well-divided database pages is obtained, and the purification of the industrial cluster information database is realized. According to the purification of industrial cluster information database, RFD algorithm is used to extract the page data features of purified industrial cluster information database. The extracted results are substituted into the information retrieval, and the vectors composed of retrieval units are used to describe the information of various types of industrial clusters and each retrieval. The matching results of information retrieval are obtained by calculating the correlation between the information of industrial clusters and the query, and the information retrieval of industrial clusters is completed. Experimental results show that the algorithm has high precision and recall ratio, short retrieval time and low energy consumption.

A retrieval algorithm of encrypted speech based on short-term cross-correlation and perceptual hashing

In order to achieve extraction perceptual features from the encryption speech as a search digest for the content-based encryption speech retrieval, we present a retrieval algorithm of encrypted speech based on short-term cross-correlation and perceptual hashing in this paper. Firstly, the study encrypts the speech file and uploads the encrypted speech data to the encryption speech database in cloud server. Secondly, the sample speech clips are obtained by the cutting operation from the speech file for scrambling encryption. The perceptual hashing sequence of the encrypted speech is constructed by extracting the short-term cross-correlation of the encrypted speech signals as the search digest. These perceptual hashing sequences are uploaded into the hashing index table of cloud server. Finally, the Hamming distance algorithm is used for the matching retrieval operation during the search. The experimental results show that the proposed algorithm of encrypted speech perceptual hashing has a better discrimination, robustness and compactness, and the perceptual hashing sequences can be extracted directly from the encrypted sample speech. Meanwhile, the encryption speech signal has high recall and precision ratios after various content preserving operations. In the whole retrieval process, the downloading and decrypting operations of speech data are not necessary.

Automatic superpixel-based segmentation method for breast ultrasound images

Automatic and accurate breast ultrasound (BUS) image segmentation is crucial to achieve effective ultrasound-based computer aided diagnosis (CAD) systems for breast cancer. However, segmenting the tumor in BUS images is often challenging due to several artifacts that degrade the quality of ultrasound images. In this study, a new two-phase method is proposed to enable automatic and accurate segmentation of BUS images by decomposing the image into superpixels with high boundary recall ratio and employing edge- and region-based information to outline the tumor. The first phase of the method obtains an initial outline of the tumor by decomposing the BUS image into coarse superpixels to enable effective estimation of their tumor likelihoods and employing a customized graph cuts algorithm to segment the superpixels. The segmentation of the superpixels is carried out using edge-based information that quantifies the image contour cue and region-based information that characterizes the texture content of the superpixels. In the second phase, the tumor outline is improved by decomposing the BUS image into fine superpixels that enable high boundary recall ratio and employing the customized graph cuts algorithm to segment the superpixel located around the initial tumor outline. Furthermore, an edge-based active contour model is used to smooth the tumor outline. The performance of the proposed method was evaluated using a database that includes 160 BUS images (86 benign and 74 malignant). The results indicate that the first phase of the proposed method was able to detect the tumor in all BUS images and obtain mean values of the true positive ratio (TPR), false positive ratio (FPR), false negative ratio (FNR), similarity ratio (SIR), Hausdorff error (HE), and mean absolute error (ME) equal to 91.68, 11.16, 8.32, 84.52, 17.59, and 4.67, respectively. In fact, the results obtained by the first phase of the proposed method outperform four existing BUS image segmentation algorithms. Moreover, the second phase of the proposed method was able to improve the tumor outlines of the first phase and achieve mean TPR, FPR, FNR, SIR, HE, and ME values of 96.04, 7.99, 3.96, 91.41, 11.66, and 3.65, respectively. These results suggest the feasibility of employing the proposed method, which enables automatic and accurate tumor segmentation in BUS images, to develop effective CAD systems for breast cancer.

MMORPG에서 GFG 쇠퇴를 위한 현금거래 구매자 탐지 방안에 관한 연구

As the rise in popularity of online games, the users start exchanging rare items for real money. As RMT (Real Money Trade) is prevalent, GFG (Gold Farming Group) who abuse RMT shows up. GFG causes social problems such as identity theft, privacy leaks. Because they needs many bot characters to gather game items. In addition, GFG induce RMT that makes in-game problems such as a destroying game economy, account hacking. Therefore, It is very important work to collapse GFG at the perspective of social and in-game. In this paper, we proposed a fundamental method for detecting RMT buyers for the collapse of GFG at the perspective of buyer by Law of Demand and Supply. We found two type of RMT by analyzing actual game data and detected RMT buyers with high recall ratio of 98% by ruled-based detection.

Gossip‐Based Resource Location Strategy in Interest Community for P2P Networks

Resource location strategy has become very popular for mass data distribution in complex and dynamic network. Searching for the objects is a fundamental problem to unstructured Peer to peer (P2P) networks. In this paper, we propose a resource location strategy for user requirements, which employs the collaborative exchange of information between peers to construct interest communities. This strategy can gather similar peers and disseminate useful information among them. Furthermore, we design an algorithm in unstructured P2P systems named Appropriate degree gossip search algorithm (ADGSA). Using this search algorithm, the performance including search success ratio, recall rate and search response time has been improved dramatically. An efficient resource location system with a fast organization of users cluster based on their requirements can be provided, preventing the creation of unliked communities. The simulation results show that our strategy has better search efficiency, short response time and high recall ratio.

Service Discovery Based on Self-Evolution of Domain Ontology and Cloud Migration of Web Service

To meet the demand of the higher precision ratio and recall ratio of service search, a method of Web service discovery is proposed based on ontology evolution and Web service migration. According to the developed cloud service platform and the existed service repository and domain ontology list, a service migration process is proposed which migrate to the cloud platform and implement unified management for Web service. This process effectively achieves that Web services migrate in and out of the cloud platform. In order to realize the self-evolution of domain ontology, an ontology alignment mechanism is designed to automatic extend and replace of ontology concept. This mechanism achieves the integration of different keywords with the same meaning and ensures the consistency of ontology concept. In this paper, the method not only enables efficient service discovery and dynamic service management, but also can achieve high concurrency support of services and supply or billing on demand. The experiments show that the proposed method solves the search problem of low efficiency based on keyword, and it effectively improves the precision ratio and recall ratio of service discovery.

Searching and Analyzing Qualitative Data on Personal Computer

The number of files stored on our personal computer (PC) is increasing very quickly and locating information in this environment is difficult. In this paper we present a set building blocks for implementing a system which is composed of four modules i.e. indexing mechanism, analyzing text, index storing and searching mechanism. The implementation of these four modules are described in details and additionally, we provide implementation of user interface, and how they interact with each other. Keywords - Desktop Search, Information Retrieval, indexing, searching, personal computer (PC). 3 I. INTRODUCTION With the development of computer technology, computer can complete many kinds of complicated tasks. Therefore, the number of files stored in the PC is increasing very quickly. At the same time, the number of various documents stored in the PC, such as digital photos, text files, video and audio files, increases in an amazing rate. However, a new problem arises; computer users have to spend much time searching the useful information in the ocean of the computer data, and sometimes, even ever seen or used files by users cannot be found. Therefore, the current problem which the users face is not how to save the file, but how to find and locate the file as quickly as possible. In other words, the traditional desktop information retrieval technology cannot meet the current needs of the users. This will inevitably lead to development of new technologies of desktop search engine. Desktop search engine is designed to help users find and locate the required information or documents from the PC effectively. Today, desktop search engine technologies become more popular in field of information retrieval. The full-text search engine is the one that can search each word in documents. The full-text search engine first indexes for each document, then search system will search the index database according to the keywords which are inputted by users. The search results will return to users according to a certain sorting algorithm. The characteristic of the search engine is huge amount of information. It segments the whole content of the document and adds to the index database so that it has high recall ratio. The full-text search engine also has the characteristics of short cycle, rapid development and cheap costs. Today there are often massive of documents stored in our PC's, we often spend much time on finding documents which are needed. The desktop search systems index the large number of documents, so that they can locate the needed documents immediately. The desktop search systems solve the problems of the difficulty of finding the right document. With the number of open-source search engine tools, we can design our personal desktop search engine conveniently and efficiently. Lucene is an open-source full-text search engine tool which is excellent and popular. The following will introduce the characteristics and the basic frameworks of Lucene. Lucene is a full-text search engine. You can search a lot of documents including specified words. The characteristics of Lucene is (8): High performance of search; High scalability of target documents; Morphological analyzer; Phrase search, regular expressions, attribute search, and similarity search; Multilingualism with Unicode; Independent of the file format and repository; Intelligent web crawler; Simple and powerful API.

Design of a P2P content recommendation system using affinity networks

The success and intensive use of social networks makes strategies for efficient document location a hot topic of research. In this paper, we propose a common vector space to describe documents and users to create a social network based on affinities, and explore epidemic routing to recommend documents according to the user’s interests. Furthermore, we propose the creation of a SoftDHT structure to improve the recommendation results. Using these mechanisms, an efficient document recommender system with a fast organization of clusters of users based on their affinity can be provided, preventing the creation of unlinked communities. We show through simulations that the proposed system has a short convergence time and presents a high recall ratio.

Detecting Bloody Images Using Multi-Features

In this paper, we present an automatic system for bloody image detection on the internet. A quick blood model in HIS color space is employed to identify the blood regions and then the fractal dimension and the information entropy of these regions are extracted. At last all the features are fed to a SVM -classifier to tell whether the image is bloody or not. Our experiments on real-world web image data indicate that this system can detect the bloody images with high recall ratio and high precision ratio.

Study of Streamline Dynamic Successive Trees Index Model for Ciphertext

Secure index is the core technology in the full-ciphertext retrieval. In order to appliance the efficient full-text retrieval in the ciphertext state, a streamline dynamic successive trees of ciphertext(SDSTC)index model is proposed. The SDSTC index model which supports queries of substring and dynamic updating of index, has a high recall and precision ratio in retrieval. Giving algorithms of model index creation, retrieval and ciphertext index updating and the analysis of its security and efficiency. Compared with other existing index models, experiments show that the SDSTC index model has good time efficiency and more suitable for areas of ciphertext security of full-text retrieval

Video-Tag Detection and Recognition

In this work, we present a video-tag detection and recognition method. According to the duration of the video, choose an appropriate strategy to sample the frames. After the candidate tag of every frame is computed, a median filter algorithm is employed to get the tag boundary. At last the binary video-tag is determined by a multi-frame-based analysis algorithm. After scaling the binary tag image to the standard size, a full image-matching algorithm is used to recognize the tag. The experimental results indicate the proposed video-tag detection method has high recall ratio and precision ratio, and the image-matching-based video-tag recognition method performs much better than the traditional OCR methods.

Lesion detection in demoscopy images with novel density-based and active contour approaches

BackgroundDermoscopy is one of the major imaging modalities used in the diagnosis of melanoma and other pigmented skin lesions. Automated assessment tools for dermoscopy images have become an important field of research mainly because of inter- and intra-observer variations in human interpretation. One of the most important steps in dermoscopy image analysis is the detection of lesion borders, since many other features, such as asymmetry, border irregularity, and abrupt border cutoff, rely on the boundary of the lesion. ResultsTo automate the process of delineating the lesions, we employed Active Contour Model (ACM) and boundary-driven density-based clustering (BD-DBSCAN) algorithms on 50 dermoscopy images, which also have ground truths to be used for quantitative comparison. We have observed that ACM and BD-DBSCAN have the same border error of 6.6% on all images. To address noisy images, BD-DBSCAN can perform better delineation than ACM. However, when used with optimum parameters, ACM outperforms BD-DBSCAN, since ACM has a higher recall ratio.ConclusionWe successfully proposed two new frameworks to delineate suspicious lesions with i) an ACM integrated approach with sharpening and ii) a fast boundary-driven density-based clustering technique. ACM shrinks a curve toward the boundary of the lesion. To guide the evolution, the model employs the exact solution [27] of a specific form of the Geometric Heat Partial Differential Equation [28]. To make ACM advance through noisy images, an improvement of the model’s boundary condition is under consideration. BD-DBSCAN improves regular density-based algorithm to select query points intelligently.

Association classification algorithm based on structure sequence in protein secondary structure prediction

Objective To propose a novel associate classification algorithm SAC (structural association classification) and develop a compound pyramid model for accurate and precise protein secondary structure prediction. Method Based on the slide window theory, the protein sequence was treated as a window with length of 13, in which the target amino acid resided in the center, while the remaining area was targeted as secondary amino acid structures. To the head and tail of the sequence, the mirror method was employed to fill the space with an opposite- position structure in relation to the central position. In the mining process, the KDD ∗ model not only focuses on the high support and confidence rules, but also pay attention to high confidence and low support rules, which is called ‘knowledge in shortage’. Towards the end of the mining process, sets H, E and C, consisted of rule sets whose consequents are α-helix, β-sheet and C-coil, were created respectively to meet the basic requirements for the protein secondary structure prediction. The knowledge base of protein secondary structure was then established with these three newly-acquired rule sets. Through the CMAR (Classification based on Multiple Association rules) algorithm, a novel multi-classifier was developed to determine the best likelihood of a given window to the secondary structure through the adjacent information on amino acid sequential window and screening of three different rule sets. Result The protein knowledge base consisted of 8049 rules corresponding to sets H, E and C with 2642, 1895 and 3512 rules, respectively, was obtained. Experiment shows, theoretically, accuracy ratio exceeded 85% when confidence threshold value was 70% and 90%. Through the classification process using the multi-classifier SAC developed in four experiments, the significantly high accuracy and recall ratios up to 83.06% (According to Q 3 criterion, followed by abbreviation) in RS126 ( Chen & Chaudhari, 2007; Guo et al., 2004; Hu et al., 2004; Liu et al., 2004) and 80.49% in CB513 ( Guo et al., 2004; Liu et al., 2004; Wang & Liu (2004)), respectively, were demonstrated. Conclusion The structural association classification algorithm with pyramid classification developed in the present study demonstrated significantly high accuracy in the protein secondary structure prediction. The study results suggest a highly reliable and accurate alternative in the contemporary protein structure prediction.

A Method of Web Information Automatic Extraction Based on XML

With the increasingly high-speed of the internet as well as the increase in the amount of data it contains, users are finding it more and more difficult to gain useful information from the web. How to extract accurate information from the Web efficiently has become an urgent problem. Web information extraction technology has emerged to solve this kind of problem. The method of Web information auto-extraction based on XML is designed through standardizing the HTML document using data translation algorism, forming an extracting rule base by learning the XPath expression of samples, and using extraction rule base to realize auto-extraction of pages of same kind. The results show that this approach should lead to a higher recall ratio and precision ratio, and the result should have a self-description, making it convenient for founding data extraction system of each domain.

Applications of Support Vector Machine Based on Boolean Kernel to Spam Filtering

Spam is so widely speared that has a bad effect on daily use of E-mail. Nowadays, among the primary technologies of spam filtering, support vector machine (SVM) is applied widely, because it is efficient and has high separating accuracy. The main problem of support vector machine arithmetic is how to choose the kernel function. To solve this problem people propose spam filtering arithmetic of support vector machine based on Boolean kernel. The arithmetic uses filtering methods based on attributes, such as IP address, subject words, keywords in content, enclosure information, etc. These attributes compose the feature vectors, and the vectors are classified by SVM-MDNF based on Boolean kernel. The experiment results show that this arithmetic has high separating accuracy, high recall ratio and precision ratio. The arithmetic has its value in theory and application.

OPAC에서 탐색결과의 클러스터링에 관한 연구

This study evaluated the applicability of the static hierarchic clustering model to clustering query results in OPAC. Two clustering methods(Between Average Linkage(BAL) and Complete Linkage(CL)) and two similarity coefficients(Dice and Jaccard) were tested on the query results retrieved from 16 title-based keyword searchings. The precision of optimal dusters was improved more than 100% compared with title-word searching. There was no difference between similarity coefficients but clustering methods in optimal cluster effectiveness. CL method is better in precision ratio but BAL is better in recall ratio at the optimal top-level and bottom-level clusters. However the differences are not significant except higher recall ratio of BAL at the top-level duster. Small number of clusters and long chain of hierarchy for optimal cluster resulted from BAL could not be desirable and efficient.

A morphology-driven string matching approach to Arabic text searching

In this paper, a morphological matching approach based on the lexical structure of Arabic words is introduced. Given a root R, the principle underlying morphological matching is to search for R in a text T to find all its occurrences. An algorithm is presented and its empirical and running time performance is analyzed. The results of searching experiments show that morphological matching provides high precision and recall ratios. But, since morphological searching is probabilistic, the results of experiments involved a number of errors caused by weak verbs. The analysis of running time performance indicates that morphological searching can be as efficient as pattern matching, if the root is entered by the user.

ＩＮＩＳ Ａｔｏｍｉｎｄｅｘ誌の利用法

Different types of manual search are presented for end users to put special features of the INIS Atomindex to practical use. High recall ratios are assured certainly from searching on either main entry section of relevant subject categories or titles under two or three groups of main headings selected systematically without any knowledge on the category scheme and INIS Thesaurus. Descriptions are also given on the frame and effective use of other four Indexes including semiannual cumulatives.

Presenting search results to meet specific user needs

Users of information services need the technical content of documents rather than the documents themselves, and the results of a literature search should ideally be presented in such form that they can be readily assimilated and applied. The degree to which information can be pre‐processed is affected by various factors including the ratio of laboratory to information staff and the context of the enquiry. The traditional bibliography may not be the best way to meet a specific information requirement, and in any case is likely to be produced by computer to an increasing extent: continuous prose may render the content of the literature more accessible. For this purpose, computerized retrieval systems should be designed to permit a fairly high recall ratio, so that the information worker can carry out a final selection of material and treat it in the way most useful to the enquirer. Some examples are given to illustrate different methods of answering enquiries.

Automatic classification and retrieval of documents by means of a bibliographic pattern discovery algorithm

We present completely automatic procedures for the self-generation of meaningful groups among the members of a document collection and for the classification according to these groups of subsequent documents. These procedures operate on large document collections with reasonably short computation times. Thus far, in our experiments on the physics literature, automatic classification has proven to be as good as or better than manual indexing and, in addition, potentially less expensive. Empirically derived, our method is based upon a pattern discovery algorithm which uses only the citation content of a document and which operates on the bibliographic links among papers. The self-generated groups correspond to very specific subject headings; retrospective bibliographies generated in the procedures allow one to classify the subsequent literature with remarkably high recall and relevance ratios, close to 100%.