Abstract: A program recommendation system employing a program record module and one of various nearest neighbor modules is disclosed. In response to a reception of a program record, the program record module converts each key field of the program record into a feature value. A single neighbor module selectively generates a recommendation of a program corresponding to the program record based upon a stored program record qualifying as a nearest neighbor of the received program record. A multiple neighbor module selectively generates a recommendation of the program corresponding to the program record based upon N number of stored program records qualifying as N number of nearest neighbors of the received program record. A neighbor cluster selectively generates a recommendation of the program corresponding to the program record based upon the cluster of stored program records qualifying as the nearest neighbor of the received program record.

Abstract: In customizing a user profile employed by a recommendation system, users are prompted for feedback regarding content that is the subject of the recommendation system. Only feedback that does not degrade performance of the recommendation system, as measured by the error rate, is accepted and utilized to modify the user profile. Feedback that would degrade performance is discarded without being employed to alter the user profile. In this manner, error is continually driven toward a minimum by system changes based on feedback.

Abstract: A search engine determines a set of other images that are similar to a user-selected image, and presents those other images to the user. In determining whether two images are sufficiently similar to each other to merit presentation of one, the search engine determines a Euclidean distance between separate feature vectors that are associated with each of the images. Each such vector indicates diverse types of information that is known about the associated image. The types of information included within such a vector may include attributes that reflect visual characteristics that are visible in an image, verbal tags that have been associated with the image users in a community of users, concepts derived from those tags, coordinates that reflect a geographic location at which a camera that produced the image was when the camera produced the image, and concepts related to groups with which the image is associated.

Abstract: A system for a meta-browser includes a receiver, an output device for displaying a virtual unified browsing space, and a selector input device for navigating the space and making selections therein. The receiver is provided with the capability to receive information from multiple sources of different media types, including broadcast programming, additional information through the broadcast, electronic program guide information, and the Internet. The meta-browser presents personalized collections of information from multiple sources of different media types as different media collections in the unified browsing space, which can be easily and intuitively browsed. A user's browsing experience is further enhanced by personalized recommendations, which are generated based on a profile of the user's interests captured across multiple media sources. Collaborative filtering, which uses multiple user profiles to compute co-relations between the user's profile and others, is also used to make recommendations.

Abstract: A television programming recommender is disclosed that selectively obtains feedback information from a user to update one or more profiles associated with the user. Previously obtained implicit and explicit preferences are utilized to selectively focus the collection of feedback information to further update and refine the implicit and explicit preferences. The present invention obtains feedback from a user in a manner that maximizes the value of the obtained information and improves the performance of the television programming recommender. The present invention automatically requests feedback from the user upon the occurrence of predefined criteria. The nature of the requested feedback, and the manner in which the obtained feedback is used to adjust a profile, can vary.

Abstract: A method is provided for reporting and analyzing user search behaviors in a large scale heterogeneous search engine platform. Content repository managers want to understand how users search for content in their repository including what categories and attributes users are interested in, how users were referred to the site, and which searchable items were viewed. The method provides a low-cost alternative to OLAP and data warehouse solutions and exploits the scalability and user interface of a search engine. Furthermore, the taxonomy of the content repository needed for analysis is already known to the search engine, and need not be exported or represented in a different format required by another tool. Data analysis can be conducted interactively and in real-time.

Abstract: Methods and systems for generating query and result-based relevance indexes are provided. For one embodiment, a plurality of queries is received from a plurality of users. Each query of the plurality of queries is measured based on one or more metrics. The measured data of each query is stored. The queries are associated into topical query categories. A performance of a first query category is calculated based on at least one metric of the one or more metrics. A metric data request for a select category is received from a client. Lastly, the stored measured data of the selected category is transmitted to the client in response to the metric data request.

Abstract: A method and apparatus are disclosed that generate television program recommendations based on queries that have previously been performed by a user on an electronic program guide. A conventional program recommender score for a given program is adjusted according to the degree of correlation between the attribute-value pairs that define the program and the attribute-value pairs that have previously been searched by the user. A historical search database indicates the number of times each attribute-value pair appears in a user query and provides additional information regarding the preferences of the user. Higher frequency counts for certain attribute-value pairs imply the user's preference for programs conforming to such criteria.

Abstract: A method and apparatus are disclosed for recommending items of interest to a user, such as television program recommendations, before a viewing history or purchase history of the user is available. A third party viewing or purchase history is processed to generate stereotype profiles that reflect the typical patterns of items selected by representative viewers. A user can select the most relevant stereotype(s) from the generated stereotype profiles and thereby initialize his or her profile with the items that are closest to his or her own interests. A clustering routine partitions the third party viewing or purchase history (the data set) into clusters, such that points (e.g., television programs) in one cluster are closer to the mean of that cluster than any other cluster. A distance computation routine evaluates the closeness of a television program to each cluster based on the distance between a given television program and the mean of a given cluster.

Abstract: A television program recommender is disclosed that automatically identifies changes in viewing preferences. Once changing viewing preferences are identified, the disclosed television programming recommender can (i) adapt the generated television program recommendations to such changes in viewing preferences, or (ii) more efficiently manage the storage of the viewing history. For cyclical or periodic changes in viewing preferences, the television programming recommender generates television program recommendations using a sub-set of the viewing history from a corresponding earlier time period. Likewise, for true or permanent changes in viewing preferences, the television programming recommender optionally generates television program recommendations using the most recent sub-set of the viewing history, which most likely reflects the current viewing preferences.

Abstract: A dynamic method and system for processing video source frames (e.g., a television movie) with a video processing system (VPS) that includes a processor, a memory structure, input devices, and an output display. The video source frames, which are received by the VPS from a video source, are executed by the processor. While the video source frames are being executed, a subset of the frames, called key frames, are extracted from the video source frames and stored in the memory structure. The extracting of the key frames is implemented in accordance with a frame extraction algorithm. The extracting is terminated prior to completion of execution of the video source frames. Following termination of extracting, the key frames are reviewed by a user with the output display to determine whether to watch the remainder of the video source frames. The key frames are erased after being reviewed.

Abstract: Methods and systems for generating query and result-based relevance indexes are provided. For one embodiment, a plurality of queries is received from a plurality of users. Each query of the plurality of queries is measured based on one or more metrics. The measured data of each query is stored. The queries are associated into topical query categories. A performance of a first query category is calculated based on at least one metric of the one or more metrics. A metric data request for a select category is received from a client. Lastly, the stored measured data of the selected category is transmitted to the client in response to the metric data request.

Abstract: An apparatus and method for recommending a schedule of events to a user is disclosed. In the preferred embodiment of the system and method, each channel schedule is broken down into time slices. A novel fuzzy-now recommendation-time value is calculated for each time slice. This fuzzy-now recommendation-time value is a two dimensional value measured in units of recommendation-time, or “enjoyment minutes”. By means of the calculated fuzzy-now recommendation-time values, recommended schedules may be generated using a wide variety of selection methods.

Abstract: Various methods and apparatuses are described for an adult content detection implementation. In one embodiment, a method detects adult content images by tracked query association to a user's query for an image search. The set of images returned in response to the user's query on a search engine are based on whether one or more images in the set are classified as an adult content image.

Abstract: A recommendation system and method are disclosed. In the system and method, the personal schedule of the user is used to modify the recommendation functions of media events. The personal schedule may be entered by the user or determined through monitoring over time. An exemplary recommendation function modification is if a media event ends after the user's bedtime, as indicated by the personal schedule. In this example, the recommendation function of that event will be reduced in value because the user will likely go to bed before the event is over.

Abstract: A system and method that locates and utilizes JavaScript national language resource files. A national language processing system for providing national language specific resources to be displayed in the Web page, including: a Web resource manager for providing a location of a JavaScript national language resource, wherein the Web resource manager includes a hash table for storing and querying previously located JavaScript national language resources; and a system for issuing an http request to a server to locate the JavaScript national language resource if the location of the JavaScript national language resource does not exist in the hash table.

Abstract: A method, system and program product for populating a user profile based on existing user profiles is provided. Specifically, under the present invention, base characteristics are designated for a new user profile. Based on the designated characteristics, the new user profile is associated with a particular Vornoi cluster region of existing user profiles. Once associated, the new user profile is populated with defined characteristics from the existing user profiles within the particular cluster region. After population, viewing recommendations can be made.

Abstract: Real-time events are electronically recommended using a fuzzy-now function of time. Real-time experiencing apparatus can be automatically tuned to a particular real-time event using a surf ring related to recommendations or automatically without a user request. Recommendations can be presented on a remote control device.

Abstract: A method and apparatus are disclosed for generating television program recommendations based on a user-specified query. The television programming recommender automatically constructs a query based on previous searches that have been executed by a user. The television programming recommender can automatically initiate a search in response to a user command using the top-N search terms for each attribute that have been previously used in a query. Each time a search is initiated by the user the query is decomposed to identify the attribute-value pairs specified by the user and a counter is incremented. When an automatic search is initiated, the top-N attribute-value pairs are retrieved for each possible attribute and automatically placed in a search bin. The attributes of each program in an electronic program guide are compared to the attribute-value pairs specified in the automatic query to identify programs satisfying the limitations of the automatic query.

Abstract: A user interface for a TV recommender system includes a display screen having a first region for displaying a rating derived from a previously defined TV viewing preference profile contained in the recommender system; and a second region displaying preference settings in the profile which were used to derive the rating. The user interface enables the preference settings to be changed if the rating derived by the profile is incorrect. Additionally, the user interface allows for new features to be added to the profile, which were not previously a part of the profile.