Data processing: database and file management or data structures – Database design – Data structure types
Reexamination Certificate
2001-07-23
2004-10-05
Coby, Frantz (Department: 2171)
Data processing: database and file management or data structures
Database design
Data structure types
C707S793000, C707S793000, C707S793000, C707S793000, C707S793000
Reexamination Certificate
active
06801909
ABSTRACT:
TECHNICAL FIELD
The present invention relates to systems and methods for searching data repositories and in particular, to a system and method for obtaining user preferences and providing user recommendations for unseen physical and information goods and services.
BACKGROUND ART
In today's information society, knowledge is often leveraged from the individual level to the community level. Distributed networks such as the Internet make it possible for information to be obtained, processed, and disseminated easily. However, as the availability of information increases, it becomes increasingly difficult for individuals to find the information they want, when they want it, and in a way that better satisfies their requirements. The overwhelming number of options and alternatives that e-commerce offers leaves many consumers confused and uncertain about what products and services may or may not meet their needs. Information seekers are able to take advantage of the myriad Internet sites purporting to provide the best and most current information on various subjects, but are unable to determine which sites are providing the most reliable and respected information. These disadvantages are especially true for consumers and information seekers who are uncertain for which physical and information goods and services they are looking. Purchasing and research decisions are increasingly more difficult when individuals are not only uninformed but also confused.
Using an approach traditionally favored by brick-and-mortar libraries, some Internet sites seek to alleviate these difficulties by serving as directories. The organizational scheme of such sites allow users to manually search for relevant documents by traversing a topic hierarchy, into which documents in a collection are categorized. The directory serves as a guide to help users reach particularly useful documents. Unfortunately, manually searching for documents online, while not as time consuming as manually searching for documents in a brick-and-mortar library, remains time consuming even if the directory is well-organized. In addition, the dynamic nature and constantly increasing size of the Internet makes it difficult to maintain current information.
The development of search engine technology and the application of such technology to the Internet has provided additional assistance to consumers and information seekers by reducing the need to manually search for documents. The first generation of Internet search engines, including Altavista, Excite/Inktomi, InfoSeek and others, incorporate the results of substantial research in information retrieval (IR) (which is concerned with the retrieval of documents from textual databases) and library science. As illustrated in
FIG. 1
, when using these search engines, a user is required to specify his/her information needs in terms of a query, which is then compared, typically at a simple keyword level, with titles of documents in a collection. The technology is designed to identify those documents that are most likely to be related to the query terms and correspondingly relevant to the user's information needs. Unfortunately, although such search engines quickly provide relevant information, they repeatedly fail to deliver accurate and current information. This is primarily because the original premises of IR, i.e., persistency of unstructured text documents and the existence of sizable collections, cannot be effectively applied to the constantly changing Internet.
Directories and search engines are considered “pull” technologies, because users must seek out the information and retrieve it from information sources. Alternate “push” technologies have been developed to reduce the time users must spend in order to obtain information from directories and sift through results returned by search engines. Users of these technologies subscribe to information “channels” and the content provider periodically “pushes” the updated information to the user's computer. From the user's point of view, the service performs automated frequent downloads of current information that is related to a user's general topic of interest. Unfortunately, this technology presents the same problems that plague the television and cable television industries—if it offers too few options, the content is very general—if it offers more options, the content is more specific, but users are unable to decide for which and how many services to subscribe. The continuous flow of information increases the need for filters, and for the user to customize those filters, to obtain information relevant enough to warrant the user's attention.
Research related to overcoming the limitations of prevalent “pull” and “push” models has recently focused on incorporating artificial intelligence (AI) technology into such models. In some modern information retrieval models, intelligent personal assistants or agents, in the form of software-based “robots” or “bots”, are instructed to seek out information, sort discovered information according to individual user preferences, and present to the user only the most relevant information. The agents seek to perform a function similar to a travel agent or a secretary. Unfortunately, even these advanced systems suffer from the need to rely on user-specified preferences. The user is typically unwilling to spend time creating a user profile. Moreover, his/her interests may change over time, which makes it difficult to maintain an accurate profile. Further, because the systems rely on the user to specify his/her preferences, the specification process is sensitive to input errors, and its effectivity depends on the user's familiarity with the business domain being searched and the functionality of the agents. These systems are also limited in that the user-specified preferences must be specific to fixed domains or applications, preventing them from being easily adapted to other domains or applications.
The most recent advanced information retrieval systems seek to learn about the user and quickly present to the user recommendations for product and information goods and services based on the learned preferences. Three examples of learning user profiles include learning by given examples, stereotyping, and observation. In systems that learn by given examples, the user is requested to answer questions or provide examples of relevant information, and the system processes the information according to internal weighting rules and builds a user profile. While the process is simple, the examples may not be representative and the results are likely to be imprecise. In systems that learn by stereotyping, the system has defined a stereotype based upon accumulated statistics and users are assigned to one of the stereotypes based upon characteristics provided by the user. While these systems require little interaction with the user, their accuracy depends on the level of detail and the number of different stereotypes, as well as the user selecting the most important personal characteristics. In systems that learn by observation, the creation of user profiles is based on first observing the user's behavior over time, including recognizing, for example search inputs and user responses to the search results, and second on matching the user's information needs and the system's actions. While these systems require no interaction with the user, and can adapt to changes in the user's interests over time, they cannot be deployed immediately because they require a training period to become effective.
These “recommender” systems use such processes to compile “meta-information”, that is, information at the knowledge level, and use several filtering layers to enable them to function more efficiently as agents and decision-making guides. Unfortunately, these first generation “recommender” systems are not effective for broad application across multiple business domains because they capture interest only in specific items. While personalizing and tailoring the recommendations to specific
Delgado Joaquin
LaPlanche Renaud
Turck Mathias
Coby Frantz
Morrison & Foerster / LLP
Triplehop Technologies, Inc.
LandOfFree
System and method for obtaining user preferences and... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System and method for obtaining user preferences and..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System and method for obtaining user preferences and... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3259905