Unlock instant, AI-driven research and patent intelligence for your innovation.

A recommendation method and system

A recommendation method and recommendation matrix technology, applied in the field of information processing, can solve problems such as high computational complexity, inability to effectively integrate and maximize utilization, and poor recommendation effect.

Active Publication Date: 2018-01-16
百度移信网络技术(北京)有限公司
View PDF11 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

At the same time, the insurmountable high computational complexity
[0004] However, there are various problems in previous recommendation methods. For example, user_based CF and item_based CF are the same algorithm; user_based CF and item_based CF have poor recommendation effects; existing similarity calculation methods are mainly based on the similarity of attribute vectors. degree calculation method, its effect is poor; the matrix decomposition algorithm is poor when adapting to implicit feedback information; the computational complexity is uncontrollable; the best solution to matrix sparsity and cold start; based on user Behavioral information and information based on content, tags, social relations, etc., which cannot be effectively integrated and maximized

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A recommendation method and system
  • A recommendation method and system
  • A recommendation method and system

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0062] First take the continuous case as an example. For the continuous case, the given weights are all 1. Examples of book recommendations in online bookstores, refer to figure 1 The similarity measurement method is described. First, as shown in step S1, the server collects all user information and all book information in the online bookstore, as well as all historical data of users clicking and reading books. Set the collection of all books in the online bookstore as the collection M (m1, m2, ...), and the collection of all users as the collection N (n1, n2, ...), assuming that in the collection M and the attribute values ​​of the elements in the set N satisfy a uniform distribution from positive infinity to negative infinity. Below we introduce how to obtain the similarity between users based on the historical data of users' operations on books without knowing any attribute information of books or users.

[0063] Now assume that the book that user n1 wants to see in use...

Embodiment 2

[0083] Taking the calculation of the similarity between users and users, or between items and items in order to recommend items to users in online shopping as an example, the comparison objects here are users and users, or items and items. refer to figure 2 Make the following instructions. First, if figure 2 As shown in step S21, the server collects information according to the user's login and registration, the items sold on the website, and the user's operation on the items, that is, the collected information includes the user, the item, and the interaction between the user and the item , to get data about users, items, and user operations on items. The server analyzes the above information, one is the user collection User, the other is the item collection Item, and the user's operation records on the items. Here, each user's operation on the item is independent of each other, and each operation expresses the same meaning, which expresses the user's interest in the item...

Embodiment 3

[0119] Embodiment 3 is an operation of enhancing similarity correlation to the results obtained in Embodiment 1. We know that the larger the variance, the more associated results, but the error also increases accordingly.

[0120] Figure 4 Show a flowchart showing the method for enhancing similarity association of embodiment 1, refer to Figure 4 Example 3 will be described. Utilize above-mentioned similarity to define formula 1, and according to the similarity result obtained in embodiment 1, in Figure 4 Step S41 is passed to any book m x and m y , with m y and m z The similarity of m y The convolution operation, as shown in Equation 9, can get m x and m z The association between books, thus expanding the scope of the association of similarity between books, enhancing the association of similarity between books, and obtaining the enhanced similarity sim(m x , m y ). Through the operation of formula 9, the variance that satisfies formula 1 also becomes 4δ 2 .

...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention relates to a recommendation method and system. The recommendation method includes the following steps: a data acquisition step, acquiring behavioral data about the user and feature data of the item; a similarity calculation step, using the acquired behavioral data and feature data to obtain a similarity matrix between the item and the item ; The recommendation matrix calculation step, using the user's behavior data to generate a probability matrix for the user to the item, and multiplying the probability matrix with the similarity matrix to obtain a recommendation matrix.

Description

technical field [0001] The invention relates to the field of information processing, in particular to a method and system for similarity measurement in the field of information processing. Background technique [0002] At present, similarity measurement is involved in many fields, and similarity analysis is performed based on various existing similarity measurement methods to make recommendations. For example, recommended methods involved in the Internet industry and other fields. [0003] Existing methods for making recommendations include the following. A system-based filtering technique generates recommendations. Including the use of existing similarity measures. Such as: Pearson correlation coefficient, Jaccard coefficient, cosine similarity, correlation similarity and other similarity measurement methods to obtain the nearest neighbors of TOP_N. Use the neighbors of TOP_N to recommend relevant results according to their operation conditions. One is to generate reco...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06F17/30
Inventor 朱宝
Owner 百度移信网络技术(北京)有限公司