Word similarity is broadly used in many applications, such as information retrieval, information extraction, text classification, word sense disambiguation, example -based machine translation, etc. There are two different methods used to compute similarity: one is based on ontology or a semantic taxonomy; the other is based on collocations of words in a corpus. As a lexical knowledgebase with rich semantic information, How-net has been employed in various researches. Unlike other thesauri, such as WordNet and Tongyici Cilin, in which word similarity is defined based on the distance between words in a semantic taxonomy tree, How-net defines a word in a complicated multi-dimensional knowledge description language. As a result, a series of problems arise in the process of word similarity computation using How-net. The difficulties are outlined below: 1. The description of each word consists of a group of sememes. For example, the Chinese word “暗箱(camera obscura)” is described as: “part|部件, #TakePicture|拍攝, %tool|用具 , body|身”, and the Chinese word “寫信 (write a letter)” is described as: “write|寫, ContentProduct=letter|信件”; 2. The meaning of a word is not a simple combination of these sememes. Sememes are organized using a specific knowledge description language. To meet these challenges, our work includes: 1. A study on the How-net knowledge description language. We rewrite the How-net definition of a word in a more structural format, using the abstract data structure of set and feature structure. 2. A study on the algorithm used to compute word similarity based on How-net. The similarity between sememes, that between sets , and that between feature structures are given. To compute the similarity between two sememes, we

/pdf/ji-yu-zhi-wang-de-ci-hui-yu-yi-xiang-si-du-ji-suan-word-3q2xf1mp2c.pdf

基於《知網》的辭彙語義相似度計算 (Word Similarity Computing Based on How-net).

Purpose – Since its introduction in 2006, messages posted to the microblogging system Twitter have provided a rich dataset for researchers, leading to the publication of over a thousand academic papers. This paper aims to identify this published work and to classify it in order to understand Twitter based research.Design/methodology/approach – Firstly the papers on Twitter were identified. Secondly, following a review of the literature, a classification of the dimensions of microblogging research was established. Thirdly, papers were qualitatively classified using open coded content analysis, based on the paper's title and abstract, in order to analyze method, subject, and approach.Findings – The majority of published work relating to Twitter concentrates on aspects of the messages sent and details of the users. A variety of methodological approaches is used across a range of identified domains.Research limitations/implications – This work reviewed the abstracts of all papers available via database search...

/pdf/what-do-people-study-when-they-study-twitter-classifying-4j1bckqno7.pdf

What do people study when they study Twitter? Classifying Twitter related academic papers

Methods and systems for identifying and presenting information are described. Search results are received from a search engine. An entity associated with a networked site is selected using a selection criterion. A determination is made as to whether the listing includes a link to the networked site or includes the link to the networked site at a list position that fails to satisfy a first criterion. After determining that the link to the site associated with the entity is not included in the search result listing, or that the link to the site associated with the entity is not at a position in the search result listing that satisfies the first criterion, the search listing is modified by causing the link to the site associated with the entity to appear in a first position in the search listing. The modified search listing is displayed on a user terminal.

Methods and systems for searching, selecting, and displaying content

Systems and methods are disclosed for monitoring hashtags in micro-blog posts, or status updates, from users to provide one or more crowd-based features. In one embodiment, a micro-blog post of a user in a crowd of users is obtained. The crowd of users is a group of users within a predefined spatial proximity to one another. A hashtag in the micro-blog post is then obtained, and a crowd-based feature is provided based on the hashtag.

Monitoring hashtags in micro-blog posts to provide one or more crowd-based features

A method and system are provided for selecting and presenting a route to a user. An exemplary embodiment includes storing a plurality of associations, each association between a route taken by a user and user characteristics. Received from a particular user is a request for at least one recommended route to a destination point. A query is formulated based on user specified parameters, including an origination point, the destination point, and characteristics of other users, to obtain routes taken by other users. A plurality of routes taken by other users having at least some of the characteristics of other users and related to at least one of the origination point and the destination point is retrieved from the stored plurality of associations. At least one recommended route is recommended to the particular user based on at least one of the retrieved routes that most closely matches the user specified parameters.

System And Method For Selecting And Presenting A Route To A User

Based on the text information processing, we have made a study on the application of support vector machine in text categorization. Through introducing the basic principle of SVM, we described the process of text classification and further proposed a SVM-based classification model. Finally, experimental data show that F1 value of SVM classifier has reached more than 86.26%, and the classification results comparing to other classification methods have greatly improved, and it also proves that SVM is an effective machine learning method.

Study on SVM Compared with the other Text Classification Methods

Text clustering is the key technology for topic detection, and topic detection is essentially similar to the unsupervised clustering. However, general clustering is based on global information, and clustering in the topic detection is based on incremental ways. So we should study topic detection according to clustering algorithm, and it is necessary for clustering algorithm to be in-depth and extensive research. Vector space model (VSM) is one of the most simple and effective topics representation model. And K-means is a well-known and widely used partitional clustering method. Therefore, we develop a topic detection prototype system to study how K in K-means affects topic detection. Then we get the variation law that it affects topic detection, and add up their optimal values in topic detection. Finally, TDT evaluation methods prove that the validity of the value of K in the algorithm is 83.33% in the topic detection prototype system based on K-means. This shows that K-means clustering algorithm is suited to deal with topic detection.

The key technology of topic detection based on K-means

Through researching and analyzing the structure of lucene package, we have developed a full-text information retrieval system on the basis of lucene full-text retrieval. After mastering index structure and principle, we increase the size of index buffer in memory and decrease the frequency of writing index to disk by a specific algorithm. What is more, we optimize index by merging it in memory and on disk. As a result, the efficiency of creating index for 70000 documents has been improved by 55.1% at the best circumstances, and that of information retrieval for 70000 documents has been improved by 15.6% at the best circumstances. Keywords-lucene; full-text retrieval; index; efficiency

Study on Efficiency of Full-Text Retrieval Based on Lucene

Text classification is the key technology for topic tracking, and vector space model (VSM) is one of the most simple and effective model for topics representation. Feature selection algorithm in VSM is an important means of data pre-processing, and it can reduce vector space dimension and improve the generalization ability of the algorithm. Therefore, it is necessary for feature selection algorithms to be in-depth and extensive research. So we study how feature space dimension and feature selection algorithm affect topic tracking. Then we get the variation law that they affect topic tracking, and add up their optimal values in topic tracking. Finally, TDT evaluation methods prove that optimal topic tracking performance based on weight of evidence for text increases by 8.762% more than mutual information.

Study on Feature Selection Algorithm in Topic Tracking

Information deficiency is a huge problem when researching on video indexing and retrieval. On the other hand, text in video frames implies lots of semantics inherently, and can provide supplemental but important information for video data processing. A smart approach for text detection, localization and extraction in video frames is presented in this paper. Here, block change rate (BCR for short) is imported to realize text detection and localization, and element image division in Lab color space is implied in binary text extraction. The approach has been implemented and tested with a variety of video sources, and the result of experiments is promising.

Shuicai Shi

Papers

Study on SVM Compared with the other Text Classification Methods

The key technology of topic detection based on K-means

Study on Efficiency of Full-Text Retrieval Based on Lucene

Study on Feature Selection Algorithm in Topic Tracking

A Smart Approach for Text Detection, Localization and Extraction in Video Frames