43results about How to "Precise personalized recommendation" patented technology

The invention provides an information recommendation system and method. The system comprises a basic data collection device, a data warehouse processing device, a basic recommendation engine device and a recommendation engine optimization device, wherein the basic data collection device is used for collecting the historical data of client browsing information and purchasing information from a financial enterprise system, wherein the historical data comprises structured data and unstructured data; the data warehouse processing device is used for receiving the historical data, converting the unstructured data into the structured data and carrying out data cleaning processing on the converted structured data and unstructured data; the basic recommendation engine device is used for calculatinga relationship between a client and a preference commodity according to the client browsing information and purchasing information in the historical data subjected to the data cleaning processing inpreset time, and obtaining a preliminary recommendation result according to the relationship between the client and the preference commodity; the recommendation engine optimization device is used forregulating the preliminary recommendation result according to a recommendation accuracy optimization regulation factor to obtain a final recommendation result, and providing the final recommendation result for a client side server. By use of the above technical scheme, information recommendation accuracy is improved.

The invention aims to provide a method and equipment for determining target object information corresponding to a user in a resource information database. Processing equipment performs match inquiring in the resource information database according to an index word of the user to obtain candidate target object information corresponding to the index word, retrieval application information corresponding to the candidate target object information is determined according to corresponding retrieval log information of the user under the index word, and the target object information corresponding to the user in the resource information database is determined according to one or more pieces of candidate target object information and the retrieval application information. Compared with the prior art, the method and the equipment have the advantages that the target object information is obtained in a specific resource range, the accuracy and completeness of the obtained target object information are improved, the target object information obtaining efficiency is improved, further services such as personalized recommendation are accurately performed for the user, and the user experience is improved.

The present invention discloses a personalized product recommendation method and system. The method includes: obtaining the historical behavior data of multiple users within a preset time period, and obtaining the first training sample after sorting according to predetermined rules; obtaining the influence factor based on the cosine similarity method as The second training sample; using the first training sample and the second training sample as the training samples of the deep learning model to train to obtain the trained deep learning model; outputting a list of commodities that the user is interested in predicted by the model. The invention effectively utilizes the timing information of commodities in the user's historical behavior, so that the commodities in the historical behavior have different weight values ​​according to the time sequence of their interactive behavior in the calculation of the recommendation system, and the commodity impact factor reflects the global characteristics of the commodity and The user's interest in the product can effectively increase the feature quantity obtained by the deep learning model, and effectively improve the personalized recommendation effect for cold-start users.

Embodiments of the invention disclose a multimedia data processing method and device. The method comprises the following steps of generating a multimedia data operation behavior matrix according to operation behaviors of a history user group for more data in a preset multimedia database; computing concealed feature vectors respectively corresponding to the multimedia data and user feature vectors respectively corresponding to history users based on a sparse self-encoding neural network according to the multimedia data operation behavior matrix; and when a recommendation request corresponding to a target user is received and the history user group includes the target user, obtaining a plurality of multimedia data in personal operation behavior information of the target user and carrying out recommendation processing on the multimedia data in the personal operation behavior information according to the user feature vector corresponding to the target user and the concealed feature vectors respectively corresponding to the multimedia data in the personal operation behavior information. Through adoption of the method and the device, the recommended song can be guaranteed to be liked by the user, so that the recommendation effect is improved.

The invention discloses an association-rule recommending method based on self-adaptive multiple minimum supports. The method comprises the following steps of firstly, establishing a commodity-classifying hierarchical tree according to commodity classification, and classifying concrete commodities according to the classifying hierarchical tree; next respectively carrying out minimum-support threshold-value setting on each concrete commodity and the upper-layer class of a concrete-commodity layer, and then mining frequent item sets and generating rules by utilizing a multiple-minimum-support association-rule expanding algorithm on the basis of the support threshold-value setting, wherein the threshold-value setting relates to the influences of time factors, concrete-commodity price factors and concrete-commodity brand factors; finally generating recommendation for each user by adopting a TOP-N recommending method. When personalized recommendation is made for the user by the association-rule recommending method, the characteristics of different objects can be better embodied by considering the influences of many factors on the multiple-minimum-support threshold-value setting for the concrete commodities and the classes; meanwhile, a data-sparsity problem and a cold-starting problem in a recommending system are relieved, so that the personalized recommendation can be more accurately made for the user.

The invention relates to a program recommendation method. The program recommendation method comprises the following steps that firstly, television programs are classified according to the types of the programs; current watching information of each user is collected at preset intervals and is used as a watching historical record of the corresponding user; accurate personalized recommendation is achieved through the social relation correlation based on the users and the content matching technology based on the historical records. By the adoption of the program recommendation method, the defects of the mode that recommendation is conducted through simple statistics of the program click rate completely based on individual interest and habit of watching the television programs are overcome, a television program provider can more accurately lock clients and improve the audience rating, and therefore accurate marketing can be achieved.