Multimodal knowledge representation method fusing entity image information and entity category information

A technology of entity image and knowledge representation, applied in the field of knowledge map reasoning, can solve problems such as data sparseness, lack of map knowledge, and low computing efficiency

Pending Publication Date: 2021-10-08
BEIJING UNIV OF POSTS & TELECOMM +1
View PDF0 Cites 6 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Network-based knowledge representation has problems such as low computational efficiency and data sparseness under large-scale knowledge graphs.
At present, most of the open knowledge graphs are mainly constructed manually or semi-automatically, which leads to the serious lack of knowledge of these graphs. At the same time, due to the storage method of the graph structure of the knowledge graph itself, the calculation efficiency is low.

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
  • Multimodal knowledge representation method fusing entity image information and entity category information
  • Multimodal knowledge representation method fusing entity image information and entity category information
  • Multimodal knowledge representation method fusing entity image information and entity category information

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0019] In order to make the above-mentioned features and advantages of the present invention more comprehensible, the present invention will be further described in detail below in conjunction with specific embodiments and accompanying drawings.

[0020] The embedding method of the entity image information designed by the present invention, wherein the entity image encoder process is as follows figure 1 as shown, figure 2 Schematic diagram of the process of constructing an entity image-based representation for the attention mechanism, the main steps of which include:

[0021] Step 101, image feature extraction. For visual knowledge, we use the VGG16 Net model pre-trained on ImageNet, and we use the vectors from the last fully connected layer as our desired image feature vectors. For image input img i , we use f i A feature vector representing an entity image.

[0022] Step 102, image feature mapping. To map entity image feature representation vectors from image space to...

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 discloses a multi-modal knowledge representation method fusing entity image information and entity category information. A model carries out united joint training on structure-based knowledge representation, image-based knowledge representation and category-based knowledge representation of entities to complete multi-modal knowledge graph representation. The method includes an entity image information embedding method which is responsible for extraction of entity image feature information and conversion from an image space to a knowledge space; according to the entity category information embedding method, modeling is carried out on a semantic relation between an entity category and a corresponding triple relation, and representation of an entity based on the category is constructed; and the multi-modal knowledge representation model fusing the entity image information and the entity category information is responsible for multi-modal knowledge representation learning fusing the entity image information and the entity category information. By constructing the multi-modal knowledge representation method, a new thought is provided for solving the problem of data sparseness in a knowledge graph reasoning technology and efficiently calculating semantic relation, fusion and reasoning performance of entities and relations.

Description

technical field [0001] The invention belongs to knowledge map reasoning technology, and in particular relates to artificial intelligence related fields such as information retrieval, question answering system and intelligent dialogue. Background technique [0002] With the advent of the Fifth Generation (5G) era, Internet technology has developed rapidly, and the volume and dimension of data have grown explosively. In the face of massive data, users are increasingly demanding more precise and intelligent searches. In order to provide users with intelligent services that can understand user needs, this problem needs to be solved urgently. Knowledge graph emerged as an intuitive knowledge expression method for discovering, managing and utilizing knowledge. To understand the knowledge map, we must first explain the knowledge base. The knowledge base (knowledge base, KB) is a knowledge system formed by structuring human knowledge, which contains basic facts, general rules and ...

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 Applications(China)
IPC IPC(8): G06F16/36G06F16/383G06F16/583G06F40/30G06N3/04G06N3/08G06N5/04
CPCG06F16/367G06F16/383G06F16/583G06F40/30G06N3/08G06N5/04G06N3/045
Inventor 刘建毅张茹李萌吕智帅
Owner BEIJING UNIV OF POSTS & TELECOMM
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products