Matches in ScholarlyData for { ?s ?p ?o. }
- 464 sameAs 464.
- 464 abstract "The aim of a web-based recommender system is to provide highly accurate and up-to-date recommendations to its users; in practice, it will hope to retain its users over time. However, this raises unique challenges. To achieve complex goals such as keeping the recommender model up-to-date over time, we need to consider a number of external requirements. Generally, these requirements arise from the physical nature of the system, for instance the available computational resources. Ideally, we would like to design a system that does not deviate from the required outcome. Modeling such a system over time requires to describe the internal dynamics as a combination of the underlying recommender model and the its users' behavior. We propose to solve this problem by applying the principles of modern control theory-a powerful set of tools to deal with dynamical systems-to construct and maintain a stable and robust recommender system for dynamically evolving environments. In particular, we introduce a design principle by focusing on the dynamic relationship between the recommender system's performance and the number of new training samples the system requires. This enables us to automate the control other external factors such as the system's update frequency. We show that, by using a Proportional-Integral-Derivative controller, a recommender system is able to automatically and accurately estimate the required input to keep the output close to a pre-defined requirements. Our experiments on a standard rating dataset show that, by using a feedback loop between system performance and training, the trade-off between the effectiveness and efficiency of the system can be well maintained. We close by discussing the widespread applicability of our approach to a variety of scenarios that recommender systems face.".
- 464 hasAuthorList authorList.
- 464 isPartOf proceedings.
- 464 keyword "control theory".
- 464 keyword "dynamical systems".
- 464 keyword "evaluation".
- 464 keyword "recommender systems".
- 464 keyword "temporal analysis".
- 464 title "Using Control Theory for Stable and Efficient Recommender Systems".
- authorList type List.
- authorList hasFirstItem item-1.
- authorList hasItem item-1.
- authorList hasItem item-2.
- authorList hasItem item-3.
- authorList hasLastItem item-3.
- item-1 type ListItem.
- item-1 hasContent tamas-jambor.
- item-1 next item-2.
- item-2 type ListItem.
- item-2 hasContent jun-wang.
- item-2 next item-3.
- item-2 previous item-1.
- item-3 type ListItem.
- item-3 hasContent neal-lathia.
- item-3 previous item-2.
- 468 creator dmitri-perelman.
- 468 creator edward-bortnikov.
- 468 creator roman-sandler.
- 468 creator ronny-lempel.
- 468 type InProceedings.
- 468 label "Lightweight Automatic Face Annotation in Media Pages".
- 468 sameAs 468.
- 468 abstract "Labeling human faces in images contained in Web media stories enables enriching the user experience offered by media sites. We propose a lightweight framework for automatic image annotation that exploits named entities mentioned in the article to significantly boost the accuracy of face recognition. While previous works in the area labor to train comprehensive offline visual models for a pre-defined universe of candidates, our approach models the people mentioned in a given story on the fly, using a standard Web image search engine as an image sampling mechanism. We overcome multiple sources of noise introduced by this ad-hoc process, to build a fast and robust end-to-end system from off-the-shelf error-prone text analysis and machine vision components. In experiments conducted on approximately 900 faces depicted in 500 stories from a major celebrity news website, we were able to correctly label $81.5\%$ of the faces while mislabeling $14.8\%$ of them.".
- 468 hasAuthorList authorList.
- 468 isPartOf proceedings.
- 468 keyword "Entity ranking".
- 468 keyword "Face detection".
- 468 keyword "Face recognition".
- 468 keyword "Image search".
- 468 keyword "Media page annotation".
- 468 title "Lightweight Automatic Face Annotation in Media Pages".
- authorList type List.
- authorList hasFirstItem item-1.
- authorList hasItem item-1.
- authorList hasItem item-2.
- authorList hasItem item-3.
- authorList hasItem item-4.
- authorList hasLastItem item-4.
- item-1 type ListItem.
- item-1 hasContent edward-bortnikov.
- item-1 next item-2.
- item-2 type ListItem.
- item-2 hasContent ronny-lempel.
- item-2 next item-3.
- item-2 previous item-1.
- item-3 type ListItem.
- item-3 hasContent dmitri-perelman.
- item-3 next item-4.
- item-3 previous item-2.
- item-4 type ListItem.
- item-4 hasContent roman-sandler.
- item-4 previous item-3.
- 475 creator nitesh-v-chawla.
- 475 creator ryan-n-lichtenwalter.
- 475 type InProceedings.
- 475 label "Vertex Collocation Profiles: Subgraph Counting for Link Analysis and Prediction".
- 475 sameAs 475.
- 475 abstract "We introduce the concept of a vertex collocation profile (VCP) for the purpose of topological link analysis and prediction. VCPs provide nearly complete information about the surrounding local structure of embedded vertex pairs. The VCP approach offers a new tool for domain experts to understand the underlying growth mechanisms in their networks and to analyze link formation mechanisms in the appropriate sociological, biological, physical, or other contexts. The same resolution that gives VCP its analytical power also enables it to perform well when used in supervised models to discriminate potential new links. We first develop the theory, mathematics, and algorithms underlying VCP. Then we demonstrate VCP methods performing link prediction competitively with unsupervised and supervised methods across several different network families. We conclude with timing results that introduce the comparative performance of several existing algorithms and the practicability of VCP computations on large networks.".
- 475 hasAuthorList authorList.
- 475 isPartOf proceedings.
- 475 keyword "Algorithms".
- 475 keyword "Performance".
- 475 keyword "Theory".
- 475 title "Vertex Collocation Profiles: Subgraph Counting for Link Analysis and Prediction".
- authorList type List.
- authorList hasFirstItem item-1.
- authorList hasItem item-1.
- authorList hasItem item-2.
- authorList hasLastItem item-2.
- item-1 type ListItem.
- item-1 hasContent ryan-n-lichtenwalter.
- item-1 next item-2.
- item-2 type ListItem.
- item-2 hasContent nitesh-v-chawla.
- item-2 previous item-1.
- 491 creator chi-wang.
- 491 creator kaushik-chakrabarti.
- 491 creator surajit-chaudhuri.
- 491 creator tao-cheng.
- 491 type InProceedings.
- 491 label "Targeted Disambiguation of Ad-hoc, Homogeneous Sets of Named Entities".
- 491 sameAs 491.
- 491 abstract "In many entity extraction applications, the entities to be recognized are constrained to be from a list of “target entities”. In many cases, these target entities are (i) ad-hoc, i.e., do not exist in a knowledge base and (ii) homogeneous (e.g., all are IT companies, all are shoe brands). We study the following novel disambiguation problem in this unique setting: given the candidate mentions of all the target entities, determine which ones are true mentions of a target entity. Prior techniques only consider target entities present in a knowledgebase and/or having a rich set of attributes. In this paper, we develop novel techniques that requires no knowledge about the entities except their names. Our main insight is to leverage the homogeneity constraint and disambiguate the candidate mentions collectively across all documents. We propose a novel graph-based model, called MentionRank, for that purpose. Furthermore, if additional knowledge is available for some or all of the entities, our model can leverage it to further improve quality. Our experiments demonstrate the effectiveness of our model. To the best of our knowledge, this is the first work on targeted entity disambiguation for ad-hoc entities.".
- 491 hasAuthorList authorList.
- 491 isPartOf proceedings.
- 491 keyword "ad-hoc entity".
- 491 keyword "graph-based ranking".
- 491 keyword "semi-supervised".
- 491 keyword "target domain".