Method for generating cross-modal representation vector and cross-modal recommendation method
A cross-modal and vector technology, applied in the field of cross-modal recommendation, can solve the problem that the model cannot be generalized, and achieve the effect of improving user experience and performance
Active Publication Date: 2021-08-24
INST OF COMPUTING TECHNOLOGY - CHINESE ACAD OF SCI
View PDF4 Cites 2 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0006] (2), the model cannot be used universally when dealing with different modalities, and it is necessary to design a data processing model for multi-modal fusion
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 moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0029] In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below through specific embodiments in conjunction with the accompanying drawings. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention.
[0030] As mentioned in the background technology section, the semantics of cross-modality in the existing technology is not precise enough, resulting in inaccurate recommendation and poor user experience. Therefore, based on entries of different modalities, the present invention uses corresponding information extraction methods to extract cross-modal consistency labels including feature vectors corresponding to the common content of different modalities and feature vectors corresponding to the exclusive content of the current modal. Based on the text information of each entry, the core...
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
Login to View More Abstract
The embodiment of the invention provides a method for generating a cross-modal representation vector and a cross-modal recommendation method. The method comprises the following steps: extracting a cross-modal consistency index comprising a feature vector corresponding to common content of different modalities and a feature vector corresponding to exclusive content of a current modal from an entry by using a corresponding information extraction mode based on entries of different modalities, and extracting a core word set based on text information of each entry; generating a cross-modal content graph containing node connections corresponding to a plurality of entries and a plurality of core words according to cross-modal consistency indexing, and generating a cross-modal representation vector by using the cross-modal content graph through a multi-modal graph neural network, so that features corresponding to cross-modal data are mapped to the same semantic space more accurately; performance of the recommendation system is improved, and user experience is improved.
Description
technical field [0001] The present invention relates to the field of cross-modal recommendation, in particular to the field of obtaining cross-modal representation vectors based on multi-modal graph neural networks for recommendation, and more specifically, to a method for generating cross-modal representation vectors and cross-modal recommendation methods. Background technique [0002] In recent years, personalized recommendation service has become an important user experience improvement service, which is widely used in picture recommendation, blog recommendation or audio, video recommendation and other aspects of network content in social network scenarios. With the great success of Internet companies such as Douyin, Kuaishou, and Toutiao in terms of the number of users or the total value of the company, personalized recommendation, especially for multi-modal data, has become a popular research point. [0003] However, the current recommendation system is often the same ...
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
Login to View More IPC IPC(8): G06F16/9535G06N3/04G06N3/08
CPCG06F16/9535G06N3/08G06N3/045
Inventor 蒋鑫龙陈益强黄武亮吴启晨
Owner INST OF COMPUTING TECHNOLOGY - CHINESE ACAD OF SCI