Text clustering is most commonly treated as a fully automated task without user feedback. However, a variety of researchers have explored mixed-initiative clustering methods which allow a user to interact with and advise the clustering algorithm. This mixed-initiative approach is especially attractive for text clustering tasks where the user is trying to organize a corpus of documents into clusters for some particular purpose (e.g., clustering their email into folders that reflect various activities in which they are involved). This paper introduces a new approach to mixed-initiative clustering that handles several natural types of user feedback. We first introduce a new probabilistic generative model for text clustering (the SpeClustering model) and show that it outperforms the commonly used mixture of multinomials clustering model, even when used in fully autonomous mode with no user input. We then describe how to incorporate four distinct types of user feedback into the clustering algorithm, and provide experimental evidence showing substantial improvements in text clustering when this user feedback is incorporated.
{"title":"Text clustering with extended user feedback","authors":"Yifen Huang, Tom Michael Mitchell","doi":"10.1145/1148170.1148242","DOIUrl":"https://doi.org/10.1145/1148170.1148242","url":null,"abstract":"Text clustering is most commonly treated as a fully automated task without user feedback. However, a variety of researchers have explored mixed-initiative clustering methods which allow a user to interact with and advise the clustering algorithm. This mixed-initiative approach is especially attractive for text clustering tasks where the user is trying to organize a corpus of documents into clusters for some particular purpose (e.g., clustering their email into folders that reflect various activities in which they are involved). This paper introduces a new approach to mixed-initiative clustering that handles several natural types of user feedback. We first introduce a new probabilistic generative model for text clustering (the SpeClustering model) and show that it outperforms the commonly used mixture of multinomials clustering model, even when used in fully autonomous mode with no user input. We then describe how to incorporate four distinct types of user feedback into the clustering algorithm, and provide experimental evidence showing substantial improvements in text clustering when this user feedback is incorporated.","PeriodicalId":433366,"journal":{"name":"Proceedings of the 29th annual international ACM SIGIR conference on Research and development in information retrieval","volume":"33 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114239169","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 novel framework for music content indexing and retrieval. The music structure information, i.e., timing, harmony and music region content, is represented by the layers of the music structure pyramid. We begin by extracting this layered structure information. We analyze the rhythm of the music and then segment the signal proportional to the inter-beat intervals. Thus, the timing information is incorporated in the segmentation process, which we call Beat Space Segmentation. To describe Harmony Events, we propose a two-layer hierarchical approach to model the music chords. We also model the progression of instrumental and vocal content as Acoustic Events. After information extraction, we propose a vector space modeling approach which uses these events as the indexing terms. In query-by-example music retrieval, a query is represented by a vector of the statistics of the n-gram events. We then propose two effective retrieval models, a hard-indexing scheme and a soft-indexing scheme. Experiments show that the vector space modeling is effective in representing the layered music information, achieving 82.5% top-5 retrieval accuracy using 15-sec music clips as the queries. The soft-indexing outperforms hard-indexing in general.
{"title":"Music structure based vector space retrieval","authors":"N. Maddage, Haizhou Li, M. Kankanhalli","doi":"10.1145/1148170.1148185","DOIUrl":"https://doi.org/10.1145/1148170.1148185","url":null,"abstract":"This paper proposes a novel framework for music content indexing and retrieval. The music structure information, i.e., timing, harmony and music region content, is represented by the layers of the music structure pyramid. We begin by extracting this layered structure information. We analyze the rhythm of the music and then segment the signal proportional to the inter-beat intervals. Thus, the timing information is incorporated in the segmentation process, which we call Beat Space Segmentation. To describe Harmony Events, we propose a two-layer hierarchical approach to model the music chords. We also model the progression of instrumental and vocal content as Acoustic Events. After information extraction, we propose a vector space modeling approach which uses these events as the indexing terms. In query-by-example music retrieval, a query is represented by a vector of the statistics of the n-gram events. We then propose two effective retrieval models, a hard-indexing scheme and a soft-indexing scheme. Experiments show that the vector space modeling is effective in representing the layered music information, achieving 82.5% top-5 retrieval accuracy using 15-sec music clips as the queries. The soft-indexing outperforms hard-indexing in general.","PeriodicalId":433366,"journal":{"name":"Proceedings of the 29th annual international ACM SIGIR conference on Research and development in information retrieval","volume":"32 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114807456","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}
The PENG system is intended to provide an integrated and personalized environment for news professionals, providing functionalities for filtering, distributed retrieval, and a flexible interface environment for the display and manipulation of news materials. In this paper we review the progress and results of the PENG system to date, and describe in detail the document filtering part of the system, which is designed to gather and filter documents to user profiles. The current architecture will be described, along with some of the main issues which have so far been found in it's development.
{"title":"PENG: integrated search of distributed news archives","authors":"Mark Baillie, F. Crestani, M. Landoni","doi":"10.1145/1148170.1148278","DOIUrl":"https://doi.org/10.1145/1148170.1148278","url":null,"abstract":"The PENG system is intended to provide an integrated and personalized environment for news professionals, providing functionalities for filtering, distributed retrieval, and a flexible interface environment for the display and manipulation of news materials. In this paper we review the progress and results of the PENG system to date, and describe in detail the document filtering part of the system, which is designed to gather and filter documents to user profiles. The current architecture will be described, along with some of the main issues which have so far been found in it's development.","PeriodicalId":433366,"journal":{"name":"Proceedings of the 29th annual international ACM SIGIR conference on Research and development in information retrieval","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131867774","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 poster, we describe the study of an interface technique that provides a list of suggested additional query terms as a searcher types a search query, in effect offering interactive query expansion (IQE) options while the query is formulated. Analysis of the results shows that offering IQE during query formulation leads to better quality initial queries, and an increased uptake of query expansion. These findings have implications for how IQE should be offered in retrieval interfaces.
{"title":"A study of real-time query expansion effectiveness","authors":"Ryen W. White, G. Marchionini","doi":"10.1145/1148170.1148332","DOIUrl":"https://doi.org/10.1145/1148170.1148332","url":null,"abstract":"In this poster, we describe the study of an interface technique that provides a list of suggested additional query terms as a searcher types a search query, in effect offering interactive query expansion (IQE) options while the query is formulated. Analysis of the results shows that offering IQE during query formulation leads to better quality initial queries, and an increased uptake of query expansion. These findings have implications for how IQE should be offered in retrieval interfaces.","PeriodicalId":433366,"journal":{"name":"Proceedings of the 29th annual international ACM SIGIR conference on Research and development in information retrieval","volume":"26 8","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132124956","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}
{"title":"A location annotation system for personal photos","authors":"Chufeng Chen, M. Oakes, J. Tait","doi":"10.1145/1148170.1148339","DOIUrl":"https://doi.org/10.1145/1148170.1148339","url":null,"abstract":"","PeriodicalId":433366,"journal":{"name":"Proceedings of the 29th annual international ACM SIGIR conference on Research and development in information retrieval","volume":"45 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132718317","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}
We describe and evaluate an approach to personalizing Web search that involves post-processing the results returned by some underlying search engine so that they re .ect the interests of a community of like-minded searchers.To do this we leverage the search experiences of the community by mining the title and snippet texts of results that have been selected by community members in response to their queries. Our approach seeks to build a community-based snippet index that re .ects the evolving interests of a group of searchers. This index is then sed to re-rank the results returned by the underlying search engine by boosting the ranking of key results that have been freq ently selected for similar q eries by community members in the past.
{"title":"Community-based snippet-indexes for pseudo-anonymous personalization in web search","authors":"Oisín Boydell, Barry Smyth","doi":"10.1145/1148170.1148283","DOIUrl":"https://doi.org/10.1145/1148170.1148283","url":null,"abstract":"We describe and evaluate an approach to personalizing Web search that involves post-processing the results returned by some underlying search engine so that they re .ect the interests of a community of like-minded searchers.To do this we leverage the search experiences of the community by mining the title and snippet texts of results that have been selected by community members in response to their queries. Our approach seeks to build a community-based snippet index that re .ects the evolving interests of a group of searchers. This index is then sed to re-rank the results returned by the underlying search engine by boosting the ranking of key results that have been freq ently selected for similar q eries by community members in the past.","PeriodicalId":433366,"journal":{"name":"Proceedings of the 29th annual international ACM SIGIR conference on Research and development in information retrieval","volume":"114 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116501286","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 Distributed Information Retrieval systems (DIR), the widely accepted solution for resource description acquisition is Query-Based Sampling (QBS) [1]. In the standard approach to QBS, once 300-500 unique documents have been retrieved sampling is curtailed. This threshold was obtained by empirically measuring the estimated resource description against the actual resource, and then considering the corresponding retrieval selection accuracy [1]. However, a fixed threshold may not generalise to other collections and environments beyond that which it was estimated on (i.e. a set of resources of uniform size [1]). Cases when the blanket application of such a heuristic would be inappropriate include (1) when the sizes of resource are highly skewed and (2) when the resources are very heterogenous. In the former, if a resource is very large then undersampling will occur because not enough documents were obtained. Conversely, if a collection is very small in size, then oversampling will occur increasing costs beyond necessity. In the later case, if the resource is varied and highly heterogeneous, then to obtain a sufficiently accurate description would require more documents to be sampled than when resources are homogenous. Either way, adopting a flat cut off will not necessarily provide sufficiently good resource descriptions for all resources.
{"title":"Adaptive query-based sampling for distributed IR","authors":"L. Azzopardi, Mark Baillie, F. Crestani","doi":"10.1145/1148170.1148277","DOIUrl":"https://doi.org/10.1145/1148170.1148277","url":null,"abstract":"In Distributed Information Retrieval systems (DIR), the widely accepted solution for resource description acquisition is Query-Based Sampling (QBS) [1]. In the standard approach to QBS, once 300-500 unique documents have been retrieved sampling is curtailed. This threshold was obtained by empirically measuring the estimated resource description against the actual resource, and then considering the corresponding retrieval selection accuracy [1]. However, a fixed threshold may not generalise to other collections and environments beyond that which it was estimated on (i.e. a set of resources of uniform size [1]). Cases when the blanket application of such a heuristic would be inappropriate include (1) when the sizes of resource are highly skewed and (2) when the resources are very heterogenous. In the former, if a resource is very large then undersampling will occur because not enough documents were obtained. Conversely, if a collection is very small in size, then oversampling will occur increasing costs beyond necessity. In the later case, if the resource is varied and highly heterogeneous, then to obtain a sufficiently accurate description would require more documents to be sampled than when resources are homogenous. Either way, adopting a flat cut off will not necessarily provide sufficiently good resource descriptions for all resources.","PeriodicalId":433366,"journal":{"name":"Proceedings of the 29th annual international ACM SIGIR conference on Research and development in information retrieval","volume":"95 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122095670","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}
Web query classification (QC) aims to classify Web users' queries, which are often short and ambiguous, into a set of target categories. QC has many applications including page ranking in Web search, targeted advertisement in response to queries, and personalization. In this paper, we present a novel approach for QC that outperforms the winning solution of the ACM KDDCUP 2005 competition, whose objective is to classify 800,000 real user queries. In our approach, we first build a bridging classifier on an intermediate taxonomy in an offline mode. This classifier is then used in an online mode to map user queries to the target categories via the above intermediate taxonomy. A major innovation is that by leveraging the similarity distribution over the intermediate taxonomy, we do not need to retrain a new classifier for each new set of target categories, and therefore the bridging classifier needs to be trained only once. In addition, we introduce category selection as a new method for narrowing down the scope of the intermediate taxonomy based on which we classify the queries. Category selection can improve both efficiency and effectiveness of the online classification. By combining our algorithm with the winning solution of KDDCUP 2005, we made an improvement by 9.7% and 3.8% in terms of precision and F1 respectively compared with the best results of KDDCUP 2005.
{"title":"Building bridges for web query classification","authors":"Dou Shen, Jian-Tao Sun, Qiang Yang, Zheng Chen","doi":"10.1145/1148170.1148196","DOIUrl":"https://doi.org/10.1145/1148170.1148196","url":null,"abstract":"Web query classification (QC) aims to classify Web users' queries, which are often short and ambiguous, into a set of target categories. QC has many applications including page ranking in Web search, targeted advertisement in response to queries, and personalization. In this paper, we present a novel approach for QC that outperforms the winning solution of the ACM KDDCUP 2005 competition, whose objective is to classify 800,000 real user queries. In our approach, we first build a bridging classifier on an intermediate taxonomy in an offline mode. This classifier is then used in an online mode to map user queries to the target categories via the above intermediate taxonomy. A major innovation is that by leveraging the similarity distribution over the intermediate taxonomy, we do not need to retrain a new classifier for each new set of target categories, and therefore the bridging classifier needs to be trained only once. In addition, we introduce category selection as a new method for narrowing down the scope of the intermediate taxonomy based on which we classify the queries. Category selection can improve both efficiency and effectiveness of the online classification. By combining our algorithm with the winning solution of KDDCUP 2005, we made an improvement by 9.7% and 3.8% in terms of precision and F1 respectively compared with the best results of KDDCUP 2005.","PeriodicalId":433366,"journal":{"name":"Proceedings of the 29th annual international ACM SIGIR conference on Research and development in information retrieval","volume":"26 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125730418","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}
Document clustering is an important tool for text analysis and is used in many different applications. We propose to incorporate prior knowledge of cluster membership for document cluster analysis and develop a novel semi-supervised document clustering model. The method models a set of documents with weighted graph in which each document is represented as a vertex, and each edge connecting a pair of vertices is weighted with the similarity value of the two corresponding documents. The prior knowledge indicates pairs of documents that known to belong to the same cluster. Then, the prior knowledge is transformed into a set of constraints. The document clustering task is accomplished by finding the best cuts of the graph under the constraints. We apply the model to the Normalized Cut method to demonstrate the idea and concept. Our experimental evaluations show that the proposed document clustering model reveals remarkable performance improvements with very limited training samples, and hence is a very effective semi-supervised classification tool.
{"title":"Document clustering with prior knowledge","authors":"Xiang-Hua Ji, W. Xu","doi":"10.1145/1148170.1148241","DOIUrl":"https://doi.org/10.1145/1148170.1148241","url":null,"abstract":"Document clustering is an important tool for text analysis and is used in many different applications. We propose to incorporate prior knowledge of cluster membership for document cluster analysis and develop a novel semi-supervised document clustering model. The method models a set of documents with weighted graph in which each document is represented as a vertex, and each edge connecting a pair of vertices is weighted with the similarity value of the two corresponding documents. The prior knowledge indicates pairs of documents that known to belong to the same cluster. Then, the prior knowledge is transformed into a set of constraints. The document clustering task is accomplished by finding the best cuts of the graph under the constraints. We apply the model to the Normalized Cut method to demonstrate the idea and concept. Our experimental evaluations show that the proposed document clustering model reveals remarkable performance improvements with very limited training samples, and hence is a very effective semi-supervised classification tool.","PeriodicalId":433366,"journal":{"name":"Proceedings of the 29th annual international ACM SIGIR conference on Research and development in information retrieval","volume":"63 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125838502","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 work tries to answer the question of what makes a query difficult. It addresses a novel model that captures the main components of a topic and the relationship between those components and topic difficulty. The three components of a topic are the textual expression describing the information need (the query or queries), the set of documents relevant to the topic (the Qrels), and the entire collection of documents. We show experimentally that topic difficulty strongly depends on the distances between these components. In the absence of knowledge about one of the model components, the model is still useful by approximating the missing component based on the other components. We demonstrate the applicability of the difficulty model for several uses such as predicting query difficulty, predicting the number of topic aspects expected to be covered by the search results, and analyzing the findability of a specific domain.
{"title":"What makes a query difficult?","authors":"David Carmel, E. Yom-Tov, Adam Darlow, D. Pelleg","doi":"10.1145/1148170.1148238","DOIUrl":"https://doi.org/10.1145/1148170.1148238","url":null,"abstract":"This work tries to answer the question of what makes a query difficult. It addresses a novel model that captures the main components of a topic and the relationship between those components and topic difficulty. The three components of a topic are the textual expression describing the information need (the query or queries), the set of documents relevant to the topic (the Qrels), and the entire collection of documents. We show experimentally that topic difficulty strongly depends on the distances between these components. In the absence of knowledge about one of the model components, the model is still useful by approximating the missing component based on the other components. We demonstrate the applicability of the difficulty model for several uses such as predicting query difficulty, predicting the number of topic aspects expected to be covered by the search results, and analyzing the findability of a specific domain.","PeriodicalId":433366,"journal":{"name":"Proceedings of the 29th annual international ACM SIGIR conference on Research and development in information retrieval","volume":"32 Suppl 4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123573442","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：