The Experts below are selected from a list of 120 Experts worldwide ranked by ideXlab platform
András A. Benczúr - One of the best experts on this subject based on the ideXlab platform.
-
MICAI (1) - Infrequent Item-to-Item Recommendation via Invariant Random Fields
Advances in Soft Computing, 2020Co-Authors: Bálint Daróczy, Frederick Ayala-gómez, András A. BenczúrAbstract:Web recommendation services bear great importance in e-commerce and social media, as they aid the user in navigating through the items that are most relevant to her needs. In a typical web site, long history of previous activities or purchases by the user is rarely available. Hence in most cases, recommenders propose items that are similar to the most recent ones viewed in the current user session. The corresponding task is called session based item-to-item recommendation. Generating item-to-item recommendations by “people who viewed this, also viewed” lists works fine for popular items. These recommender systems rely on item-to-item similarities and item-to-item transitions for building next-item recommendations. However, the performance of these methods deteriorates for rare (i.e., infrequent) items with short transaction history. Another difficulty is the cold-start problem, items that recently appeared and had no time yet to accumulate a sufficient number of transactions. In this paper, we describe a probabilistic similarity model based on Random Fields to approximate item-to-item transition probabilities. We give a generative model for the item interactions based on arbitrary distance measures over the items including explicit, implicit ratings and External Metadata. We reach significant gains in particular for recommending items that follow rare items. Our experiments on various publicly available data sets show that our new model outperforms both simple similarity baseline methods and recent item-to-item recommenders, under several different performance metrics.
-
Session Recommendation via Recurrent Neural Networks over Fisher Embedding Vectors.
Sensors, 2019Co-Authors: Domokos Kelen, Bálint Daróczy, Frederick Ayala-gómez, Anna Ország, András A. BenczúrAbstract:Recommendation services bear great importance in e-commerce, shopping, tourism, and social media, as they aid the user in navigating through the items that are most relevant to their needs. In order to build recommender systems, organizations log the item consumption in their user sessions by using different sensors. For instance, Web sites use Web data loggers, museums and shopping centers rely on user in-door positioning systems to register user movement, and Location-Based Social Networks use Global Positioning System for out-door user tracking. Most organizations do not have a detailed history of previous activities or purchases by the user. Hence, in most cases recommenders propose items that are similar to the most recent ones viewed in the current user session. The corresponding task is called session based, and when only the last item is considered, it is referred to as item-to-item recommendation. A natural way of building next-item recommendations relies on item-to-item similarities and item-to-item transitions in the form of “people who viewed this, also viewed” lists. Such methods, however, depend on local information for the given item pairs, which can result in unstable results for items with short transaction history, especially in connection with the cold-start items that recently appeared and had no time yet to accumulate a sufficient number of transactions. In this paper, we give new algorithms by defining a global probabilistic similarity model of all the items based on Random Fields. We give a generative model for the item interactions based on arbitrary distance measures over the items, including explicit, implicit ratings and External Metadata to estimate and predict item-to-item transition probabilities. We exploit our new model in two different item similarity algorithms, as well as a feature representation in a recurrent neural network based recommender. Our experiments on various publicly available data sets show that our new model outperforms simple similarity baseline methods and combines well with recent item-to-item and deep learning recommenders under several different performance metrics.
-
Item-to-item recommendation based on Contextual Fisher Information.
arXiv: Information Retrieval, 2016Co-Authors: Bálint Daróczy, Frederick Ayala-gómez, András A. BenczúrAbstract:Web recommendation services bear great importance in e-commerce, as they aid the user in navigating through the items that are most relevant to her needs. In a typical Web site, long history of previous activities or purchases by the user is rarely available. Hence in most cases, recommenders propose items that are similar to the most recent ones viewed in the current user session. The corresponding task is called session based item-to-item recommendation. For frequent items, it is easy to present item-to-item recommendations by "people who viewed this, also viewed" lists. However, most of the items belong to the long tail, where previous actions are sparsely available. Another difficulty is the so-called cold start problem, when the item has recently appeared and had no time yet to accumulate sufficient number of transactions. In order to recommend a next item in a session in sparse or cold start situations, we also have to incorporate item similarity models. In this paper we describe a probabilistic similarity model based on Random Fields to approximate item-to-item transition probabilities. We give a generative model for the item interactions based on arbitrary distance measures over the items including explicit, implicit ratings and External Metadata. The model may change in time to fit better recent events and recommend the next item based on the updated Fisher Information. Our new model outperforms both simple similarity baseline methods and recent item-to-item recommenders, under several different performance metrics and publicly available data sets. We reach significant gains in particular for recommending a new item following a rare item.
David Pease - One of the best experts on this subject based on the ideXlab platform.
-
Seamlessly integrating disk and tape in a multi-tiered distributed file system
2015 IEEE 31st International Conference on Data Engineering, 2015Co-Authors: Ioannis Koltsidas, Slavisa Sarafijanovic, Martin Petermann, Nils Haustein, Harald Seipp, Robert Haas, Jens Jelitto, Thomas Weigold, Edwin Childers, David PeaseAbstract:The explosion of data volumes in enterprise environments and limited budgets have triggered the need for multi-tiered storage systems. With the bulk of the data being extremely infrequently accessed, tape is a natural fit for storing such data. In this paper we present our approach to a file storage system that seamlessly integrates disk and tape, enabling a bottomless and cost-effective storage architecture that can scale to accommodate Big Data requirements. The proposed system offers access to data through a POSIX filesystem interface under a single global namespace, optimizing the placement of data across disk and tape tiers. Using a self-contained, standardized and open filesystem format on the removable tape media, the proposed system avoids dependence on proprietary software and External Metadata servers to access the data stored on tape. By internally managing the tape tier resources, such as tape drives and cartridges, the system relieves the user from the burden of dealing with the complexities of tape storage. Our implementation, which is based on the GPFS and LTFS filesystems, demonstrates the applicability of the proposed architecture in real-world environments. Our experimental evaluation has shown that this is a very promising approach in terms scalability, performance and manageability. The proposed system has been productized by IBM as LTFS Enterprise Edition.
-
ICDE - Seamlessly integrating disk and tape in a multi-tiered distributed file system
2015 IEEE 31st International Conference on Data Engineering, 2015Co-Authors: Ioannis Koltsidas, Slavisa Sarafijanovic, Martin Petermann, Nils Haustein, Harald Seipp, Robert Haas, Jens Jelitto, Thomas Weigold, Edwin Childers, David PeaseAbstract:The explosion of data volumes in enterprise environments and limited budgets have triggered the need for multi-tiered storage systems. With the bulk of the data being extremely infrequently accessed, tape is a natural fit for storing such data. In this paper we present our approach to a file storage system that seamlessly integrates disk and tape, enabling a bottomless and cost-effective storage architecture that can scale to accommodate Big Data requirements. The proposed system offers access to data through a POSIX filesystem interface under a single global namespace, optimizing the placement of data across disk and tape tiers. Using a self-contained, standardized and open filesystem format on the removable tape media, the proposed system avoids dependence on proprietary software and External Metadata servers to access the data stored on tape. By internally managing the tape tier resources, such as tape drives and cartridges, the system relieves the user from the burden of dealing with the complexities of tape storage. Our implementation, which is based on the GPFS and LTFS filesystems, demonstrates the applicability of the proposed architecture in real-world environments. Our experimental evaluation has shown that this is a very promising approach in terms scalability, performance and manageability. The proposed system has been productized by IBM as LTFS Enterprise Edition.
Chieko Asakawa - One of the best experts on this subject based on the ideXlab platform.
-
ASSETS - Providing synthesized audio description for online videos
Proceeding of the eleventh international ACM SIGACCESS conference on Computers and accessibility - ASSETS '09, 2009Co-Authors: Masatomo Kobayashi, Hironobu Takagi, Kentarou Fukuda, Chieko AsakawaAbstract:We describe an initial attempt to develop a common platform for adding an audio description (AD) to an online video so that blind and visually impaired people can enjoy such material. A speech synthesis technology allows content providers to offer the AD at minimal cost. We exploit External Metadata so that the AD can be independent of the video format. The External approach also allows External supporters to add ADs to any online videos. Our technology includes an authoring tool for writing AD scripts, a Web browser add-on for synthesizing ADs synchronized with original videos, and a text-based format to exchange AD scripts.
-
Accessibility commons: a Metadata repository for web accessibility
ACM SIGWEB Newsletter, 2009Co-Authors: S Kawanaka, Hironobu Takagi, Masatomo Kobayashi, Chieko AsakawaAbstract:Metadata plays a crucial role for creating a more accessible Web environment, spanning from simple alternative texts for images to various types of UI widgets used in dynamic content. Beyond the current accessibility improvement methods that rely on inlined Metadata embedded by the developers, various active research projects are focusing on External Metadata which is the key to improve even existing content without touching the original code, so Web accessibility can be more flexible and adaptive to various user needs. Accessibility Commons is a new proposal to build a standardized shared repository for Externalized Metadata. It was initially proposed by IBM and three universities last year, and then proved its applicability through the Social Accessibility project. Now is the time to move beyond single-system-implementation to a federated data infrastructure for a wide range of Web accessibility research activities and a more adaptive Web environment for people throughout the world. After introducing the basic ideas, research projects, and technical challenges, we will discuss future directions, the potential, and broader applications.
-
ASSETS - Social accessibility: achieving accessibility through collaborative Metadata authoring
Proceedings of the 10th international ACM SIGACCESS conference on Computers and accessibility - Assets '08, 2008Co-Authors: Hironobu Takagi, S Kawanaka, Masatomo Kobayashi, Takashi Itoh, Chieko AsakawaAbstract:Web content is under the control of site owners, and therefore the site owners have the responsibility to make their content accessible. This is a basic assumption of Web accessibility. Users who want access to inaccessible content must ask the site owners for help. However, the process is slow and too often the need is mooted before the content becomes accessible. Social Accessibility is an approach to drastically reduce the burden on site owners and to shorten the time to provide accessible Web content by allowing volunteers worldwide to - renovate' any webpage on the Internet. Users encountering Web access problems anywhere at any time will be able to immediately report the problems to a social computing service. Volunteers can be quickly notified, and they can easily respond by creating and publishing the requested accessibility Metadata--also helping any other users who encounter the same problems. Site owners can learn about the methods for future accessibility renovations based on the volunteers' External Metadata. There are two key technologies to enable this process, the External Metadata that allows volunteers to annotate existing Web content, and the social computing service that supports the collaborative renovations. In this paper, we will first review previous approaches, and then propose the Social Accessibility approach. The scenario, implementation, and results of a pilot service are introduced, followed by discussion of future directions.
-
ASSETS - Aibrowser for multimedia: introducing multimedia content accessibility for visually impaired users
Proceedings of the 9th international ACM SIGACCESS conference on Computers and accessibility - Assets '07, 2007Co-Authors: Hisashi Miyashita, Hironobu Takagi, Daisuke Sato, Chieko AsakawaAbstract:Multimedia content with Rich Internet Applications using Dynamic HTML (DHTML) and Adobe Flash is now becoming popular in various websites. However, visually impaired users cannot deal with such content due to audio interference with the speech from screen readers and intricate structures strongly optimized for sighted users. We have been developing an Accessibility Internet Browserfor Multimedia (aiBrowser) to address these problems. Thebrowser has two novel features: non-visual multimedia audiocontrols and alternative user interfaces using ExternalMetadata. First, by using the aiBrowser, users can directlycontrol the audio from the embedded media with fixed shortcutkeys. Therefore, this allows blind users to increase ordecrease the media volume, and pause or stop the mediato handle conflicts between the audio of the media and thespeech from the screen reader. Second, the aiBrowser canprovide an alternative simplified user interface suitable forscreen readers by using External Metadata, which can evenbe applied to dynamic content such as DHTML and Flash. In this paper, we discuss accessibility problems with multimedia content due to streaming media and the dynamic changes in such content, and explain how the aiBrowser addresses these problems by describing non-visual multimedia audio controls and External Metadata-based alternative user interfaces. The evaluation of the aiBrowser was conducted by comparing it to JAWS, one of the most popular screen readers, on three well known multimedia-content-intensive websites. The evaluation showed that the aiBrowser made the contentthat was inaccessible with JAWS relatively accessibleby using the multimedia audio controls and alternative interfaceswith Metadata which included alternative text, headinginformation, and so on. It also drastically reduced thekeystrokes for navigation with aiBrowser, which implies toimprove the non-visual usability.
Ioannis Koltsidas - One of the best experts on this subject based on the ideXlab platform.
-
Seamlessly integrating disk and tape in a multi-tiered distributed file system
2015 IEEE 31st International Conference on Data Engineering, 2015Co-Authors: Ioannis Koltsidas, Slavisa Sarafijanovic, Martin Petermann, Nils Haustein, Harald Seipp, Robert Haas, Jens Jelitto, Thomas Weigold, Edwin Childers, David PeaseAbstract:The explosion of data volumes in enterprise environments and limited budgets have triggered the need for multi-tiered storage systems. With the bulk of the data being extremely infrequently accessed, tape is a natural fit for storing such data. In this paper we present our approach to a file storage system that seamlessly integrates disk and tape, enabling a bottomless and cost-effective storage architecture that can scale to accommodate Big Data requirements. The proposed system offers access to data through a POSIX filesystem interface under a single global namespace, optimizing the placement of data across disk and tape tiers. Using a self-contained, standardized and open filesystem format on the removable tape media, the proposed system avoids dependence on proprietary software and External Metadata servers to access the data stored on tape. By internally managing the tape tier resources, such as tape drives and cartridges, the system relieves the user from the burden of dealing with the complexities of tape storage. Our implementation, which is based on the GPFS and LTFS filesystems, demonstrates the applicability of the proposed architecture in real-world environments. Our experimental evaluation has shown that this is a very promising approach in terms scalability, performance and manageability. The proposed system has been productized by IBM as LTFS Enterprise Edition.
-
ICDE - Seamlessly integrating disk and tape in a multi-tiered distributed file system
2015 IEEE 31st International Conference on Data Engineering, 2015Co-Authors: Ioannis Koltsidas, Slavisa Sarafijanovic, Martin Petermann, Nils Haustein, Harald Seipp, Robert Haas, Jens Jelitto, Thomas Weigold, Edwin Childers, David PeaseAbstract:The explosion of data volumes in enterprise environments and limited budgets have triggered the need for multi-tiered storage systems. With the bulk of the data being extremely infrequently accessed, tape is a natural fit for storing such data. In this paper we present our approach to a file storage system that seamlessly integrates disk and tape, enabling a bottomless and cost-effective storage architecture that can scale to accommodate Big Data requirements. The proposed system offers access to data through a POSIX filesystem interface under a single global namespace, optimizing the placement of data across disk and tape tiers. Using a self-contained, standardized and open filesystem format on the removable tape media, the proposed system avoids dependence on proprietary software and External Metadata servers to access the data stored on tape. By internally managing the tape tier resources, such as tape drives and cartridges, the system relieves the user from the burden of dealing with the complexities of tape storage. Our implementation, which is based on the GPFS and LTFS filesystems, demonstrates the applicability of the proposed architecture in real-world environments. Our experimental evaluation has shown that this is a very promising approach in terms scalability, performance and manageability. The proposed system has been productized by IBM as LTFS Enterprise Edition.
Barbara Schreur - One of the best experts on this subject based on the ideXlab platform.
-
accessing web based multimedia contents for the visually challenged combined tree structure and xml Metadata
SCSS (1), 2010Co-Authors: Victoria Christy Sathya Rajasekar, Barbara SchreurAbstract:We present an Integrated Multimedia Interface for the VIsually Challenged (IMIVIC), a multimedia data visualization technique for a web browser that combines a tree structure and XML Metadata. We discuss two different approaches to provide an efficient multimedia interface for the blind people, which turned out to be important in the construction of IMIVIC. The first approach is the implementation of web contents with a tree structure, which integrates many different web pages into a single page. The second approach is the creation of External Metadata to describe dynamically changing multimedia contents. We demonstrate our approach in an on-line training website and show how this combined approach helps visually challenged users access multimedia data on the web, with appropriate test.
-
SCSS (1) - Accessing Web Based Multimedia Contents for the Visually Challenged: Combined Tree Structure and XML Metadata
Innovations and Advances in Computer Sciences and Engineering, 2009Co-Authors: Victoria Christy Sathya Rajasekar, Barbara SchreurAbstract:We present an Integrated Multimedia Interface for the VIsually Challenged (IMIVIC), a multimedia data visualization technique for a web browser that combines a tree structure and XML Metadata. We discuss two different approaches to provide an efficient multimedia interface for the blind people, which turned out to be important in the construction of IMIVIC. The first approach is the implementation of web contents with a tree structure, which integrates many different web pages into a single page. The second approach is the creation of External Metadata to describe dynamically changing multimedia contents. We demonstrate our approach in an on-line training website and show how this combined approach helps visually challenged users access multimedia data on the web, with appropriate test.
