Traditional question answering systems extract answers in terms of the relevance between answer and question. However, when there are multiple relevant answers to a question, people usually use other dimensions besides relevance to select answers. Abstraction level is a frequently used dimension to distinguish general answer and specific answer. Previous question answering systems seldom consider the dimension. This paper proposes a way to calculate the abstraction level of answer. Experiments show that abstraction level can improve question answering in certain situations.
{"title":"Using Abstraction Level in Question Answering System","authors":"Bei Xu, Xiaodong Wang, H. Zhuge","doi":"10.1109/SKG.2018.00040","DOIUrl":"https://doi.org/10.1109/SKG.2018.00040","url":null,"abstract":"Traditional question answering systems extract answers in terms of the relevance between answer and question. However, when there are multiple relevant answers to a question, people usually use other dimensions besides relevance to select answers. Abstraction level is a frequently used dimension to distinguish general answer and specific answer. Previous question answering systems seldom consider the dimension. This paper proposes a way to calculate the abstraction level of answer. Experiments show that abstraction level can improve question answering in certain situations.","PeriodicalId":265760,"journal":{"name":"2018 14th International Conference on Semantics, Knowledge and Grids (SKG)","volume":"73 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121466092","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In this paper, we proposes a system for extracting multi-granular phases for event evolutions on social network platforms like Sina Weibo and Twitter. Existing studies on event extraction usually use a set of tweets to describe an event, which is not able to present the evolutional knowledge about the event. In many decision-making scenarios, it is much helpful to detect the evolutional stage of an event, as this can help people make counter-measures according to the current developing trend of the event. In this paper, we present a multi-granular approach for extracting the phases of evolutional events. We implement a web-based prototype called MGP (Multi-Granular Phase) which can extract and show the stages of events from a fine granularity such as hour to a coarse granularity like month. After a brief introduction on the architecture of MGP, we present the implemental details of MGP. Then, we present a case study to demonstrate the usability and effectiveness of MGP.
{"title":"MGP: Extracting Multi-Granular Phases for Evolutional Events on Social Network Platforms","authors":"Jialing Liang, Lin Mu, Peiquan Jin","doi":"10.1109/SKG.2018.00046","DOIUrl":"https://doi.org/10.1109/SKG.2018.00046","url":null,"abstract":"In this paper, we proposes a system for extracting multi-granular phases for event evolutions on social network platforms like Sina Weibo and Twitter. Existing studies on event extraction usually use a set of tweets to describe an event, which is not able to present the evolutional knowledge about the event. In many decision-making scenarios, it is much helpful to detect the evolutional stage of an event, as this can help people make counter-measures according to the current developing trend of the event. In this paper, we present a multi-granular approach for extracting the phases of evolutional events. We implement a web-based prototype called MGP (Multi-Granular Phase) which can extract and show the stages of events from a fine granularity such as hour to a coarse granularity like month. After a brief introduction on the architecture of MGP, we present the implemental details of MGP. Then, we present a case study to demonstrate the usability and effectiveness of MGP.","PeriodicalId":265760,"journal":{"name":"2018 14th International Conference on Semantics, Knowledge and Grids (SKG)","volume":"133 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115775810","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Management and decision-making of science and technology in the era of big data needs capabilities of data mining and analysis on massive data, simulation of complex systems, identification and prediction of potential risks and opportunities of science and technology development. China's scientific and technological innovation map (CSTIM) comprehensively utilizes information technologies including text analysis, information organization, data mining, and visualization to meet the requirements of science and technology management and decision-making in China. It realizes the dynamic, interactive, and visual display of China's scientific and technological innovation information. It could be used to analyze different levels of scientific and technological innovation such as nation, region, province, city, institution and individual researcher. It aims to help users to discover the laws of scientific and technological innovation, predict the possible trends, and support national, regional, provincial, municipal and innovative agencies' management and decision-making of science and technology.
{"title":"Mapping Science and Technology Innovation of China","authors":"Junsheng Zhang, Zhaofeng Zhang, Yan Yang, Deshan Xu, Changqing Yao, Zhihui Liu, Cheng Dong","doi":"10.1109/SKG.2018.00021","DOIUrl":"https://doi.org/10.1109/SKG.2018.00021","url":null,"abstract":"Management and decision-making of science and technology in the era of big data needs capabilities of data mining and analysis on massive data, simulation of complex systems, identification and prediction of potential risks and opportunities of science and technology development. China's scientific and technological innovation map (CSTIM) comprehensively utilizes information technologies including text analysis, information organization, data mining, and visualization to meet the requirements of science and technology management and decision-making in China. It realizes the dynamic, interactive, and visual display of China's scientific and technological innovation information. It could be used to analyze different levels of scientific and technological innovation such as nation, region, province, city, institution and individual researcher. It aims to help users to discover the laws of scientific and technological innovation, predict the possible trends, and support national, regional, provincial, municipal and innovative agencies' management and decision-making of science and technology.","PeriodicalId":265760,"journal":{"name":"2018 14th International Conference on Semantics, Knowledge and Grids (SKG)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130517293","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In recent years, Massive Open Online Courses (MOOCs) raise wide concern of the in academia. Researchers are working on making MOOCs more efficient and easier to learn. A number of works focus on describing the characters of learners via their behaviours to personalize. Unfortunately, few studies pay attention to construct the curricula system of MOOCs. While reasonable curricula system can also improve the learning efficiency remarkably. To improve the reasonability of the curricula system of MOOCs, this paper crawls all the 112920 reviews from Coursera.org (up to Jun./30/2017) for the first time, and from which we investigate the relationship between courses, learners and job markets for the purpose of discovering any helpful suggestions. The contributions of this paper include three aspects: Firstly, it discovered the topological graph of the courses through analyzing learners' reviews. Secondly, the tendency in the number of reviews per course is found for fitting power-law distribution ideally. And which perhaps means most learners only concerns very few courses of the MOOCs. Thirdly, comparing with the data from the job markets, we have some useful suggestion. In addition, the tends in the number of reviews over time are also identified. It is a key role for the time distribution of the reviews in this study. Furthermore, some effective suggestion for enhancing levels of activity in courses is presented in this paper.
{"title":"Improve the Curricula System of MOOCs Via Data Miningape","authors":"Mingming Zhao, Zhiyi Chen, Min Li","doi":"10.1109/SKG.2018.00011","DOIUrl":"https://doi.org/10.1109/SKG.2018.00011","url":null,"abstract":"In recent years, Massive Open Online Courses (MOOCs) raise wide concern of the in academia. Researchers are working on making MOOCs more efficient and easier to learn. A number of works focus on describing the characters of learners via their behaviours to personalize. Unfortunately, few studies pay attention to construct the curricula system of MOOCs. While reasonable curricula system can also improve the learning efficiency remarkably. To improve the reasonability of the curricula system of MOOCs, this paper crawls all the 112920 reviews from Coursera.org (up to Jun./30/2017) for the first time, and from which we investigate the relationship between courses, learners and job markets for the purpose of discovering any helpful suggestions. The contributions of this paper include three aspects: Firstly, it discovered the topological graph of the courses through analyzing learners' reviews. Secondly, the tendency in the number of reviews per course is found for fitting power-law distribution ideally. And which perhaps means most learners only concerns very few courses of the MOOCs. Thirdly, comparing with the data from the job markets, we have some useful suggestion. In addition, the tends in the number of reviews over time are also identified. It is a key role for the time distribution of the reviews in this study. Furthermore, some effective suggestion for enhancing levels of activity in courses is presented in this paper.","PeriodicalId":265760,"journal":{"name":"2018 14th International Conference on Semantics, Knowledge and Grids (SKG)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134318454","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Unstructured knowledge extraction is the process of recognizing and storing valuable knowledge from the natural language texts. However, few tools are available to automatically extract knowledge concepts and their relations from the text books, especially for those in Chinese. This paper proposed a method to implement the ‘example of’ and ‘part of’ semantic relations' and their related entities' extracting from the digital textbooks in Chinese. The experimental data shows that the extraction of the both relations and the entities can achieve a rather high accuracy and satisfied results comparing with the previous studies.
{"title":"Extraction and Application of Cognitive Related Semantic Relationships","authors":"Qinge Wang, Xiaofang Kuang, Weiwei Yan, Juan Yang","doi":"10.1109/SKG.2018.00039","DOIUrl":"https://doi.org/10.1109/SKG.2018.00039","url":null,"abstract":"Unstructured knowledge extraction is the process of recognizing and storing valuable knowledge from the natural language texts. However, few tools are available to automatically extract knowledge concepts and their relations from the text books, especially for those in Chinese. This paper proposed a method to implement the ‘example of’ and ‘part of’ semantic relations' and their related entities' extracting from the digital textbooks in Chinese. The experimental data shows that the extraction of the both relations and the entities can achieve a rather high accuracy and satisfied results comparing with the previous studies.","PeriodicalId":265760,"journal":{"name":"2018 14th International Conference on Semantics, Knowledge and Grids (SKG)","volume":"119 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115598183","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper first presents results of our three recent research projects on using social network analysis (SNA) techniques to analyze business big data involving stock data, trading data, and business contract data. The analysis on historical stock data identifies alternative representative indexing stock groups. The analysis on high frequency trading data establishes new algorithms for more effective high frequency trading. The analysis on business contract networks studies relationships between companies' contracts and their performance in profits and stock levels. The paper then discusses approaches to incorporating explicit semantics into conventional social networks and extending standard social network analysis techniques to more effective semantics-based analysis.
{"title":"Toward Semantic Social Network Analysis for Business Big Data","authors":"W. Du","doi":"10.1109/SKG.2018.00050","DOIUrl":"https://doi.org/10.1109/SKG.2018.00050","url":null,"abstract":"This paper first presents results of our three recent research projects on using social network analysis (SNA) techniques to analyze business big data involving stock data, trading data, and business contract data. The analysis on historical stock data identifies alternative representative indexing stock groups. The analysis on high frequency trading data establishes new algorithms for more effective high frequency trading. The analysis on business contract networks studies relationships between companies' contracts and their performance in profits and stock levels. The paper then discusses approaches to incorporating explicit semantics into conventional social networks and extending standard social network analysis techniques to more effective semantics-based analysis.","PeriodicalId":265760,"journal":{"name":"2018 14th International Conference on Semantics, Knowledge and Grids (SKG)","volume":"49 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127241569","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Hongtao Liu, Lulu Guo, Long Chen, Xueyan Liu, Zhenjia Zhu
The core of the recommendation system is the recommendation algorithm, especially the application of collaborative filtering recommendation algorithm is the most widely used. With the rapid increase of data sparsity. This paper aims at the problem of data sparsity in collaborative filtering algorithms. By mining the hidden information behind the user and the project, that is, considering different factors in the user's personal rating habits, and using Cosine and Jaccard to calculate the full degree of similarity to effectively use the rate data, improves the similarity calculation method, and solves the problem of low accuracy of the recommendation due to inaccuracy of similarity calculation. This is more in line with the logic of real life and can produce reasonable recommendations.
{"title":"Effective Similarity Measures of Collaborative Filtering Recommendations Based on User Ratings Habits","authors":"Hongtao Liu, Lulu Guo, Long Chen, Xueyan Liu, Zhenjia Zhu","doi":"10.1109/SKG.2018.00026","DOIUrl":"https://doi.org/10.1109/SKG.2018.00026","url":null,"abstract":"The core of the recommendation system is the recommendation algorithm, especially the application of collaborative filtering recommendation algorithm is the most widely used. With the rapid increase of data sparsity. This paper aims at the problem of data sparsity in collaborative filtering algorithms. By mining the hidden information behind the user and the project, that is, considering different factors in the user's personal rating habits, and using Cosine and Jaccard to calculate the full degree of similarity to effectively use the rate data, improves the similarity calculation method, and solves the problem of low accuracy of the recommendation due to inaccuracy of similarity calculation. This is more in line with the logic of real life and can produce reasonable recommendations.","PeriodicalId":265760,"journal":{"name":"2018 14th International Conference on Semantics, Knowledge and Grids (SKG)","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125525984","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
With the fast growing of social media used in our society, user role mining, as one of the most important research domains of social media analysis, attracts more and more researchers' attention. Its research results can be applied to all walks of life, e.g., recommendation system, viral marketing, etc. Lots of researchers have presented many methods to mine user roles. However, most of the existing methods just analyse the user influence rather than mine user role. Therefore, user behaviour network based user role mining method of web event is proposed. User behaviour network is firstly built. Four network topologies (e.g., degree centrality, closeness centrality, betweenness centrality, and eigenvector centrality) are used as the basis to measure users, and combining number of comment, number of repost, and statistical characteristic to mine three different user roles (information producer, information driver, and information bridger) of web event. Experimental results on the Weibo datasets show the effectiveness of the proposed model.
{"title":"User Behaviour Network Based User Role Mining of Web Event","authors":"Q. Ma, Xiangfeng Luo, Mingming Zhao","doi":"10.1109/SKG.2018.00038","DOIUrl":"https://doi.org/10.1109/SKG.2018.00038","url":null,"abstract":"With the fast growing of social media used in our society, user role mining, as one of the most important research domains of social media analysis, attracts more and more researchers' attention. Its research results can be applied to all walks of life, e.g., recommendation system, viral marketing, etc. Lots of researchers have presented many methods to mine user roles. However, most of the existing methods just analyse the user influence rather than mine user role. Therefore, user behaviour network based user role mining method of web event is proposed. User behaviour network is firstly built. Four network topologies (e.g., degree centrality, closeness centrality, betweenness centrality, and eigenvector centrality) are used as the basis to measure users, and combining number of comment, number of repost, and statistical characteristic to mine three different user roles (information producer, information driver, and information bridger) of web event. Experimental results on the Weibo datasets show the effectiveness of the proposed model.","PeriodicalId":265760,"journal":{"name":"2018 14th International Conference on Semantics, Knowledge and Grids (SKG)","volume":"46 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124717271","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Different readers are usually interested in different aspects of an article. One aspect of content may be distributed at different locations within an article. Furthermore, users often wish to obtain extended content for further understanding. Most recommendation systems recommend one or multiple pieces of texts instead of content. It is time-consuming for readers to find the required content in recommended texts or extended content from other texts. This paper designs a system that can help reader to quickly browse the content that meets users' personalized interests on one aspect or one topic. The system takes a User Interest Model as input, and retrieves the extension of the reading content from the references or other articles that quotes this article. Experimental system demonstrates that the reading system has potential for quickly reading a long article.
{"title":"The Implementation of a Personalized Reading System","authors":"Jiade Chen","doi":"10.1109/SKG.2018.00048","DOIUrl":"https://doi.org/10.1109/SKG.2018.00048","url":null,"abstract":"Different readers are usually interested in different aspects of an article. One aspect of content may be distributed at different locations within an article. Furthermore, users often wish to obtain extended content for further understanding. Most recommendation systems recommend one or multiple pieces of texts instead of content. It is time-consuming for readers to find the required content in recommended texts or extended content from other texts. This paper designs a system that can help reader to quickly browse the content that meets users' personalized interests on one aspect or one topic. The system takes a User Interest Model as input, and retrieves the extension of the reading content from the references or other articles that quotes this article. Experimental system demonstrates that the reading system has potential for quickly reading a long article.","PeriodicalId":265760,"journal":{"name":"2018 14th International Conference on Semantics, Knowledge and Grids (SKG)","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133787251","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper proposes a genetic algorithm (GA) based error correcting output codes (ECOC) algorithms. In our algorithm, some randomly initialized coding matrices are generated as seeds firstly, and our algorithm produces optimal coding matrices based on them in the evolutionary process. In our GA, each gene stands for an action, indicating two selected columns and an operator. The operators are proposed to generate new columns by exchanging information between a pair of parent columns. In this way, each individual represents a new coding matrix. A legality checking function is embedded in the GA to keep the produced coding matrix both legal and effective. At the end of this evolutionary process, the best coding matrix is selected as final solution. The experimental results show that our algorithm can efficiently optimize the coding matrix compared with the seed matrices.
{"title":"A Novel Genetic Algorithm Based ECOC Algorithm","authors":"Xiao-Na Ye, Kun-hong Liu","doi":"10.1109/SKG.2018.00030","DOIUrl":"https://doi.org/10.1109/SKG.2018.00030","url":null,"abstract":"This paper proposes a genetic algorithm (GA) based error correcting output codes (ECOC) algorithms. In our algorithm, some randomly initialized coding matrices are generated as seeds firstly, and our algorithm produces optimal coding matrices based on them in the evolutionary process. In our GA, each gene stands for an action, indicating two selected columns and an operator. The operators are proposed to generate new columns by exchanging information between a pair of parent columns. In this way, each individual represents a new coding matrix. A legality checking function is embedded in the GA to keep the produced coding matrix both legal and effective. At the end of this evolutionary process, the best coding matrix is selected as final solution. The experimental results show that our algorithm can efficiently optimize the coding matrix compared with the seed matrices.","PeriodicalId":265760,"journal":{"name":"2018 14th International Conference on Semantics, Knowledge and Grids (SKG)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123525054","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：