一种个性化移动搜索技术的研究
|
摘要
今天,Internet已进入更先进的以用户为中心的理念竞争阶段,只有最大化满足用户的需求才能产生更大的经济效益。为用户提供更贴心、更具个性化的服务,逐渐成为Internet应用研究的热点；此外,移动应用作为一种可以随时随地获取信息的途径,受到了越来越多用户的青睐,而移动搜索更是成为移动应用的最常用功能。相对于传统的桌面搜索,移动搜索具有屏幕小、查准率要求高等特点。只有满足用户查询信息的个性化需求,返回最贴近用户真实需要的结果,才能从根本上提高用户的满意度。但是,当前移动搜索技术仅仅是简单的将互联网搜索移植到移动终端,而没有相应的个性化解决方案。本文将个性化服务与移动搜索结合起来,提出了一个卓有成效的个性化移动搜索解决方案。本文所做的主要工作如下：
首先,在介绍移动搜索个性化需求的基础上,深入探讨了移动搜索的相关理论与个性化服务的各项技术,研究了它们的优点以及不足。
其次,在有用户兴趣建模技术的基础上,将扩散激活模型引入到移动用户兴趣模型的更新过程中,提出了一种基于用户相关反馈的兴趣模型建立与更新方法,通过在概念树中不断扩散激活因子,并调整概念的特征向量来不断更新每个概念的兴趣度。
之后,给出了一种基于显隐式结合的用户兴趣信息获取方法,首先在用户注册时隐式确定分类的兴趣度,之后根据用户浏览过程中的浏览行为与浏览内容提出一种获取用户的兴趣页面的方法；针对获取用户兴趣模型之后的排序,给出了一种基于概念与查询相似度的个性化重排名方法。
最后,详细介绍了采用Lucene实现的原型系统,通过实验进行对比,证明本文所设计的系统在Top-n查准率性能上与Google相比有了较大的提升。
本文所研究的内容有效分析了移动个性化搜索的实际问题,对移动信息检索的设计与实现具有一定的参考价值。所设计的个性化移动搜索系统是一种有效的探索,具有较广泛的研究意义和应用价值。
Nowadays, WWW has entered the more advanced phase of focusing on users' demand. Only by capturing users' real interest and intention and meeting users' demand better can we make more economic benefits. It has been a hot research topic to supply users'with a heart-to-heart and more personalized service. Besides, with the rapid development and popularization of mobile wireless internet, mobile wireless ap-plication has attracted more and more users for its capability and convenience of gett-ing information anytime and anywhere. Mobile search has been the most popular and useful function among these applications. In contrast to conventional search engines, mobile search has the characteristics of smaller screen, higher demand on precise rate. Only by meeting the query user's demand of personalization can we provide users wit-h the most proper result and improve the satisfactory quality. Based on the traits abov-e, this article proposes an effective solution to personalization of mobile search by co-mbining personalized service and mobile search technology. The main task of this art-icle is as follows:
Firstly, deeply studies every kind of related theory of mobile search and persona-lization service on the basis of detailedly introducing the personalization demand of mobile wireless search, and discusses their advantages and disadvantages.
Secondly, introduces the spreading activation model into the course of updating user profile and designs a method to build and update user profile, which is based on ODP category systems and relevance feedback. It can regulate every interest score in ODP hiberarchy and every concept's term vector by spreading activation score into e-very concept continually.
Thirdly, proposes a method of capturing user's individuality information by impl-ictly and explicitly grasping user's intention and behavior, which decides every categ-ory's interest score when users register, then propose a method to get interesting pages on the basis of the surfing behavior and surfing content; aimed at re-ranking the result after capturing user profile introduce a personalized re-rank method based the similar-ity between concept and query.
Lastly, detailedly introduces a prototype system, which is implemented by Lucen-e indexer, then designs and implements corresponding experiments. After the campar-ision of the performance, it has been approved that our designed system has a modera-te improvement in Top-n precise rate.
The research content of this article effectively analyzes the actual problem in mobile wireless personalized search, and has reference value for the design and impl-entation of mobile information retrieval in a certain extent. The designed personalized mobile search is an effective exploration and has a comprehensive research significa-nce and application value.
首先,在介绍移动搜索个性化需求的基础上,深入探讨了移动搜索的相关理论与个性化服务的各项技术,研究了它们的优点以及不足。
其次,在有用户兴趣建模技术的基础上,将扩散激活模型引入到移动用户兴趣模型的更新过程中,提出了一种基于用户相关反馈的兴趣模型建立与更新方法,通过在概念树中不断扩散激活因子,并调整概念的特征向量来不断更新每个概念的兴趣度。
之后,给出了一种基于显隐式结合的用户兴趣信息获取方法,首先在用户注册时隐式确定分类的兴趣度,之后根据用户浏览过程中的浏览行为与浏览内容提出一种获取用户的兴趣页面的方法；针对获取用户兴趣模型之后的排序,给出了一种基于概念与查询相似度的个性化重排名方法。
最后,详细介绍了采用Lucene实现的原型系统,通过实验进行对比,证明本文所设计的系统在Top-n查准率性能上与Google相比有了较大的提升。
本文所研究的内容有效分析了移动个性化搜索的实际问题,对移动信息检索的设计与实现具有一定的参考价值。所设计的个性化移动搜索系统是一种有效的探索,具有较广泛的研究意义和应用价值。
Nowadays, WWW has entered the more advanced phase of focusing on users' demand. Only by capturing users' real interest and intention and meeting users' demand better can we make more economic benefits. It has been a hot research topic to supply users'with a heart-to-heart and more personalized service. Besides, with the rapid development and popularization of mobile wireless internet, mobile wireless ap-plication has attracted more and more users for its capability and convenience of gett-ing information anytime and anywhere. Mobile search has been the most popular and useful function among these applications. In contrast to conventional search engines, mobile search has the characteristics of smaller screen, higher demand on precise rate. Only by meeting the query user's demand of personalization can we provide users wit-h the most proper result and improve the satisfactory quality. Based on the traits abov-e, this article proposes an effective solution to personalization of mobile search by co-mbining personalized service and mobile search technology. The main task of this art-icle is as follows:
Firstly, deeply studies every kind of related theory of mobile search and persona-lization service on the basis of detailedly introducing the personalization demand of mobile wireless search, and discusses their advantages and disadvantages.
Secondly, introduces the spreading activation model into the course of updating user profile and designs a method to build and update user profile, which is based on ODP category systems and relevance feedback. It can regulate every interest score in ODP hiberarchy and every concept's term vector by spreading activation score into e-very concept continually.
Thirdly, proposes a method of capturing user's individuality information by impl-ictly and explicitly grasping user's intention and behavior, which decides every categ-ory's interest score when users register, then propose a method to get interesting pages on the basis of the surfing behavior and surfing content; aimed at re-ranking the result after capturing user profile introduce a personalized re-rank method based the similar-ity between concept and query.
Lastly, detailedly introduces a prototype system, which is implemented by Lucen-e indexer, then designs and implements corresponding experiments. After the campar-ision of the performance, it has been approved that our designed system has a modera-te improvement in Top-n precise rate.
The research content of this article effectively analyzes the actual problem in mobile wireless personalized search, and has reference value for the design and impl-entation of mobile information retrieval in a certain extent. The designed personalized mobile search is an effective exploration and has a comprehensive research significa-nce and application value.
引文
[1]徐萍,田启树.网络信息资源的特点及其对信息用户的影响.湖南农业大学学报,2007,8(04)：108-110
[2]Jay L.关于搜索引擎的统计调查.http://www.se-express.com/9238/comment10. htm,2008-04-17
[3]Kamvar M, Baluja S. A Large Scale Study of Wireless Search Behavior:Google Mobile Search. In:Proc of the ACM SIGCHI Conference, Human Factors in Co-mputing Systems. Washington,2006,701-709
[4]William T, Harding, Anita J, et al. Cookies and Web bugs:What They Are and H-ow They Work Together. Information System Management,2001,18(3):17-24
[5]李勇,徐振宁,张维明.Internet个性化信息服务研究综述.计算机工程与应用,2002,12(19)：183-188
[6]赵志荣.个性化搜索引擎的研究、设计与实现：[四川大学硕士论文].成都：四川大学信息与档案管理系,2002,13-22
[7]Lewis D. Representation and Learning in Information Retrieval, [dissertation]. University of Massachusetts,1992,89-97
[8]Salton G, Lesk M E. Computer Evaluation of Indexing and Text Processing. In:P-roc of the ACM SIGCHI Conf. Washington:1968,8-36
[9]Salton G. The SMART Retrieval System:Experiments in Automatic Document P-rocessing:Prentice Hall Inc,1971,43-51
[10]林华兵.移动搜索中的查询优化和主题分类策略：[福州大学硕士学位论文].福州：福建大学计算机系,2006,23-46
[11]姚全珠,张楠,杨增辉等.基于压缩后缀数组技术的搜索引擎.计算机工程,2008,34(8)：83-85
[12]李小明,闫宏飞,王继明.搜索引擎、原理技术与系统.北京：科学出版社,2004,30-50
[13]吴丽辉,王斌,张刚.一个个性化的Web信息采集模型.计算机工程,2005,31(3)：86-88
[14]江涛樊,孝忠.主题爬虫的设计与实现.计算机应用,2008,11(6)：270-272
[15]Miller R C, Bharat K. SPHINX:A Framework for Creating Personal, Site-specif-ic Web Crawlers. In:Proc of the 7th International WWW Conference. Brisbance, 1998,21-30
[16]Kamba T, Bharat K, Albers M C. The Krakatoa Chronicle-An Interactive, Person-alized Newspaper on the Web. In:Proc of the Fourth International World Wide Web Conference. Boston,1995,159-170
[17]Page L, Brin S, Motwani R et al. The PageRank Citation Ranking:Bringing Ord-er to the Web. http://ilpubs.stanford.edu:8090/422.1998
[18]Zhu J H, Hong J, Hughes J. Page Rate:Counting Web Users'Votes. In:Proc of th-e 12th ACM Conference on Hypertext and Hypermedia. Denmark,2001,131-132
[19]Oztekin B U, Ertoz L, Kumar V. Usage Aware PageRank. In:Proc of the World Wide Web Conference,2005,219-230
[20]Scapira A. Collaboratively Searching the Web-An Initial Study Center for Intelli-gent Information Retrieval:[dissertation]. Massachusetts:University of Massach-usetts,1999,20-48
[21]Byong T Z, Yong W S. Personalized Web-Document Filtering using Reinforcem-ent Learning. Applied Artificial Intelligence,2001,15(7):665-685
[22]Richardson M, Domingos P. The Intelligent Surfer:Probabliistic Combination of Link and Content Information in PageRank. Cambridge:MIT Press,2002,1441-1 448
[23]Haveliwala T H. Topic-Sensitive PageRank. In:Proc of the 11th International World Wide Web Conference,2002,45-49
[24]Arasu A, Novak J, Tomkins A, et al. PageRank Computation and the Structure of the Web:Experiments and Algorithms. In:Proc of the 11th International World Wide Web Conference. ACM,2002,230-235
[25]Broder A, Kumar R, Maghoul F et al. Graph Structure in the Web. In:Proc of the 9th International World Wide Web Conference. Hungary,2000,12-15
[26]Kamvar S D, Haveliwala T H, Manning C D, et al. Extrapolation Methods for Ac-celerating PageRank Computations. In:Proc of the 12th International World Wide Web Conference. Budapest,2003,261-270
[27]Chen L,Sycara K. WebMate:A Personal Agent for Browsing and Searching. In: Proc of the 2nd International Conference on Autonomous Agents. Minnesota,199 8,132-139
[28]Chau M, Zeng D, Chen H. Personalized Spiders for Web Search and Analysis. In: Proc of the 1st ACM/IEEE Joint Conference on Digital Libraries.Virginia,2001, 79-87
[29]Scime A, Kerschberg L. WebSifter:An Ontology-based Personalizable Search A-gent for the web. In:Proc of International Conference on Digital Libraries:Rese- aries Research and Practice, Kyoto,2000,203-210
[30]吴晓,李丹宁,林洁等.个性化搜索引擎中用户兴趣模型的研究.第三届全国信息检索与内容安全学术会议.苏州,2007,32-35
[31]赵银春,付关友,朱征宇.基于Web浏览内容和行为相结合的用户兴趣挖掘.计算机工程,2005,31(12)：93-94
[32]田萱,孟祥光.智能信息检索中个性化模型的表示形式研究.情报学报,2008,23(1)：21-26
[33]李勇.智能检索中基于本体的个性化用户建模技术及应用：[国防科学技术大学硕士学位论文].湖南：国防科学技术大学计算机学院,2002,35-40
[34]邓爱林.电子商务推荐系统关键技术研究：[复旦大学博士学位论文].上海：复旦大学计算机与信息技术系,2003,41-47
[35]苏雪峰.隐交互式个性化搜索引擎研究：[太原理工大学硕士论文].山西：太原理工大学计算机与软件学院,2008,32-37
[36]Pazzani M, Muramatsu J, Billsus D. Syskill & Webert:Identifying Interesting W-eb Sites. In:Proc of the 13th National Conference on Artificial Intelligence. Port-land,1996,54-61
[37]Asnicar F, Tasso C. IfWeb:A Prototype of User Model-based Intelligent Agent f-or Documentation Filtering and Navigation in the World Wide Web. In:Proc of the 6th International Conference on User Modeling. Sardinia,1997,3-12
[38]Pazzani M, Billsus D. Learning and Revising User profiles:The Identification of Interesting Web Sites, Machine Learning,1997,27(3):313-331
[39]Adomavicius G, Tuzhilin A. Using Data Mining Methods to Build Customer Pro-files, IEEE Computer,2001,74-82
[40]Schwab I, Kobsa A, Koychev I. Learning about User from Oberservation. In:Pro of AAAI Spring Symposiumon Adaptive User Interface,2000,102-107
[41]林鸿飞,杨元生.用户兴趣模型的表示和更新机制.计算机研究与发展,200239(7)：843-847
[42]张卫丰,徐宝文.基于WWW缓冲的用户实时二维兴趣模型.计算机学报,2004,27(4)：461-470
[43]Rumelhart D, Norman D. Representation in Memory. Technical Report Center for Human Information Processing. University of California,1983,38-40
[44]Preece S. A Spreading Activation Model for Information Retrieval. [dissertation], Univ. of Illinois,1981,85-96
[45]Collins, Allan M, Loftus, Elizabeth F. A Spreading Activation Theory of Semanti-c Processing. Psychological Review,1975,82(6):407-428
[46]Sieg A, Mobasher B, Burke R. Ontological User Profiles for Personalized Web S-earch. In:Proc of the 5th Workshop on Intelligent Techniques for Web Personali-zation, Vancouver,2007,84-91
[47]严华云,刘其平,肖良军.信息检索中的相关反馈技术综述.计算机应用研究,2009,26(1)：11-14
[48]陈玉娥.个性化服务中用户模型的研究与设计：[山东科技大学硕士学位论文].山东：山东科技大学信息科学与工程学院,2007,20-25
[49]William T, Harding, Anita J, et al. Cookies and Web bugs:What They Are and How They Work Together. Information System Management,2001, (9):17-24
[50]Belkin N, Cool C, Kelly D, et al. Query Length in Interactive Information Retrieval. SIGIR 03:205-212
[51]李爱明.个性化搜索引擎用户模型研究：[华中师范大学硕士学位论文].湖北：华中师范大学,2007,32-36
[52]应晓敏.面向Internet个性化服务的用户建模技术研究：[中国人民解放军国防科学技术大学博士学位论文].湖南：中国人民解放军国防科学技术大学,2003,21-28
[53]李晓明.天网搜索.http://e.pku.edu.cn.2004-04-12
[54]Tanudjaja F, Mui L. Persona:A Contextualized and Personalized Web Search. In: the 35th Annual Hawaii International Conference on System Science,2002, 32-41
[55]Otis Gospodnetic, Erik Hatcher. Lucene In Action中文版(谭鸿,黎俊鸿等),电子工业出版社.2007,101-323
[56]Cleverdon C. The Cranfield Tests on Index Language Devices. Readings in Infor-mation Retrieval,1997,23-42
[2]Jay L.关于搜索引擎的统计调查.http://www.se-express.com/9238/comment10. htm,2008-04-17
[3]Kamvar M, Baluja S. A Large Scale Study of Wireless Search Behavior:Google Mobile Search. In:Proc of the ACM SIGCHI Conference, Human Factors in Co-mputing Systems. Washington,2006,701-709
[4]William T, Harding, Anita J, et al. Cookies and Web bugs:What They Are and H-ow They Work Together. Information System Management,2001,18(3):17-24
[5]李勇,徐振宁,张维明.Internet个性化信息服务研究综述.计算机工程与应用,2002,12(19)：183-188
[6]赵志荣.个性化搜索引擎的研究、设计与实现：[四川大学硕士论文].成都：四川大学信息与档案管理系,2002,13-22
[7]Lewis D. Representation and Learning in Information Retrieval, [dissertation]. University of Massachusetts,1992,89-97
[8]Salton G, Lesk M E. Computer Evaluation of Indexing and Text Processing. In:P-roc of the ACM SIGCHI Conf. Washington:1968,8-36
[9]Salton G. The SMART Retrieval System:Experiments in Automatic Document P-rocessing:Prentice Hall Inc,1971,43-51
[10]林华兵.移动搜索中的查询优化和主题分类策略：[福州大学硕士学位论文].福州：福建大学计算机系,2006,23-46
[11]姚全珠,张楠,杨增辉等.基于压缩后缀数组技术的搜索引擎.计算机工程,2008,34(8)：83-85
[12]李小明,闫宏飞,王继明.搜索引擎、原理技术与系统.北京：科学出版社,2004,30-50
[13]吴丽辉,王斌,张刚.一个个性化的Web信息采集模型.计算机工程,2005,31(3)：86-88
[14]江涛樊,孝忠.主题爬虫的设计与实现.计算机应用,2008,11(6)：270-272
[15]Miller R C, Bharat K. SPHINX:A Framework for Creating Personal, Site-specif-ic Web Crawlers. In:Proc of the 7th International WWW Conference. Brisbance, 1998,21-30
[16]Kamba T, Bharat K, Albers M C. The Krakatoa Chronicle-An Interactive, Person-alized Newspaper on the Web. In:Proc of the Fourth International World Wide Web Conference. Boston,1995,159-170
[17]Page L, Brin S, Motwani R et al. The PageRank Citation Ranking:Bringing Ord-er to the Web. http://ilpubs.stanford.edu:8090/422.1998
[18]Zhu J H, Hong J, Hughes J. Page Rate:Counting Web Users'Votes. In:Proc of th-e 12th ACM Conference on Hypertext and Hypermedia. Denmark,2001,131-132
[19]Oztekin B U, Ertoz L, Kumar V. Usage Aware PageRank. In:Proc of the World Wide Web Conference,2005,219-230
[20]Scapira A. Collaboratively Searching the Web-An Initial Study Center for Intelli-gent Information Retrieval:[dissertation]. Massachusetts:University of Massach-usetts,1999,20-48
[21]Byong T Z, Yong W S. Personalized Web-Document Filtering using Reinforcem-ent Learning. Applied Artificial Intelligence,2001,15(7):665-685
[22]Richardson M, Domingos P. The Intelligent Surfer:Probabliistic Combination of Link and Content Information in PageRank. Cambridge:MIT Press,2002,1441-1 448
[23]Haveliwala T H. Topic-Sensitive PageRank. In:Proc of the 11th International World Wide Web Conference,2002,45-49
[24]Arasu A, Novak J, Tomkins A, et al. PageRank Computation and the Structure of the Web:Experiments and Algorithms. In:Proc of the 11th International World Wide Web Conference. ACM,2002,230-235
[25]Broder A, Kumar R, Maghoul F et al. Graph Structure in the Web. In:Proc of the 9th International World Wide Web Conference. Hungary,2000,12-15
[26]Kamvar S D, Haveliwala T H, Manning C D, et al. Extrapolation Methods for Ac-celerating PageRank Computations. In:Proc of the 12th International World Wide Web Conference. Budapest,2003,261-270
[27]Chen L,Sycara K. WebMate:A Personal Agent for Browsing and Searching. In: Proc of the 2nd International Conference on Autonomous Agents. Minnesota,199 8,132-139
[28]Chau M, Zeng D, Chen H. Personalized Spiders for Web Search and Analysis. In: Proc of the 1st ACM/IEEE Joint Conference on Digital Libraries.Virginia,2001, 79-87
[29]Scime A, Kerschberg L. WebSifter:An Ontology-based Personalizable Search A-gent for the web. In:Proc of International Conference on Digital Libraries:Rese- aries Research and Practice, Kyoto,2000,203-210
[30]吴晓,李丹宁,林洁等.个性化搜索引擎中用户兴趣模型的研究.第三届全国信息检索与内容安全学术会议.苏州,2007,32-35
[31]赵银春,付关友,朱征宇.基于Web浏览内容和行为相结合的用户兴趣挖掘.计算机工程,2005,31(12)：93-94
[32]田萱,孟祥光.智能信息检索中个性化模型的表示形式研究.情报学报,2008,23(1)：21-26
[33]李勇.智能检索中基于本体的个性化用户建模技术及应用：[国防科学技术大学硕士学位论文].湖南：国防科学技术大学计算机学院,2002,35-40
[34]邓爱林.电子商务推荐系统关键技术研究：[复旦大学博士学位论文].上海：复旦大学计算机与信息技术系,2003,41-47
[35]苏雪峰.隐交互式个性化搜索引擎研究：[太原理工大学硕士论文].山西：太原理工大学计算机与软件学院,2008,32-37
[36]Pazzani M, Muramatsu J, Billsus D. Syskill & Webert:Identifying Interesting W-eb Sites. In:Proc of the 13th National Conference on Artificial Intelligence. Port-land,1996,54-61
[37]Asnicar F, Tasso C. IfWeb:A Prototype of User Model-based Intelligent Agent f-or Documentation Filtering and Navigation in the World Wide Web. In:Proc of the 6th International Conference on User Modeling. Sardinia,1997,3-12
[38]Pazzani M, Billsus D. Learning and Revising User profiles:The Identification of Interesting Web Sites, Machine Learning,1997,27(3):313-331
[39]Adomavicius G, Tuzhilin A. Using Data Mining Methods to Build Customer Pro-files, IEEE Computer,2001,74-82
[40]Schwab I, Kobsa A, Koychev I. Learning about User from Oberservation. In:Pro of AAAI Spring Symposiumon Adaptive User Interface,2000,102-107
[41]林鸿飞,杨元生.用户兴趣模型的表示和更新机制.计算机研究与发展,200239(7)：843-847
[42]张卫丰,徐宝文.基于WWW缓冲的用户实时二维兴趣模型.计算机学报,2004,27(4)：461-470
[43]Rumelhart D, Norman D. Representation in Memory. Technical Report Center for Human Information Processing. University of California,1983,38-40
[44]Preece S. A Spreading Activation Model for Information Retrieval. [dissertation], Univ. of Illinois,1981,85-96
[45]Collins, Allan M, Loftus, Elizabeth F. A Spreading Activation Theory of Semanti-c Processing. Psychological Review,1975,82(6):407-428
[46]Sieg A, Mobasher B, Burke R. Ontological User Profiles for Personalized Web S-earch. In:Proc of the 5th Workshop on Intelligent Techniques for Web Personali-zation, Vancouver,2007,84-91
[47]严华云,刘其平,肖良军.信息检索中的相关反馈技术综述.计算机应用研究,2009,26(1)：11-14
[48]陈玉娥.个性化服务中用户模型的研究与设计：[山东科技大学硕士学位论文].山东：山东科技大学信息科学与工程学院,2007,20-25
[49]William T, Harding, Anita J, et al. Cookies and Web bugs:What They Are and How They Work Together. Information System Management,2001, (9):17-24
[50]Belkin N, Cool C, Kelly D, et al. Query Length in Interactive Information Retrieval. SIGIR 03:205-212
[51]李爱明.个性化搜索引擎用户模型研究：[华中师范大学硕士学位论文].湖北：华中师范大学,2007,32-36
[52]应晓敏.面向Internet个性化服务的用户建模技术研究：[中国人民解放军国防科学技术大学博士学位论文].湖南：中国人民解放军国防科学技术大学,2003,21-28
[53]李晓明.天网搜索.http://e.pku.edu.cn.2004-04-12
[54]Tanudjaja F, Mui L. Persona:A Contextualized and Personalized Web Search. In: the 35th Annual Hawaii International Conference on System Science,2002, 32-41
[55]Otis Gospodnetic, Erik Hatcher. Lucene In Action中文版(谭鸿,黎俊鸿等),电子工业出版社.2007,101-323
[56]Cleverdon C. The Cranfield Tests on Index Language Devices. Readings in Infor-mation Retrieval,1997,23-42