Multi-modal information fusion broadcast television news keyword and abstract combined extraction method
A radio and television and keyword technology, which is applied in the field of multimodal information fusion radio and television news keywords and abstracts joint extraction, can solve the problem that the grammatical features and syntactic structure features of keywords cannot be correctly captured, and the pictographic character features of Chinese characters cannot be described. Affect the keyword extraction effect and other issues to achieve the effect of improving the extraction effect
Active Publication Date: 2021-11-02
CHENGDU SOBEY DIGITAL TECH CO LTD
View PDF10 Cites 1 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Existing methods relying on traditional methods and basic neural network models cannot correctly capture the grammatical and syntactic features of keywords
Relying solely on text information cannot describe the unique pictographic character features of Chinese characters, thus affecting the extraction effect of keywords
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
[0082] like figure 1 As shown, this embodiment proposes a multi-modal information fusion radio and television news keyword and abstract joint extraction method, including the following steps:
[0083] A. Collect radio and television news texts, obtain keywords and abstracts from radio and television news texts, and use the obtained keywords and abstracts to construct a joint extraction data set of radio and television news keywords and abstracts;
[0084] B, build the matt font image bank that comprises Chinese character, English character and Chinese-English punctuation;
[0085] C. Using the pre-training model and attention mechanism, based on deep learning technology to fuse image and text semantic information, build a multi-modal information fusion radio and television news keyword and abstract joint extraction model, and adopt a multi-task learning method to use radio and television news A multi-modal information fusion radio and television news keyword and abstract join...
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 invention provides a multi-modal information fusion broadcast television news keyword and abstract combined extraction method, which comprises the following steps: A, collecting broadcast television news texts, obtaining keywords and abstracts, and constructing a broadcast television news keyword and abstract combined extraction data set; B, constructing a matts font image library; C, constructing and training a multi-modal information fusion broadcast television news keyword and abstract combined extraction model; and D, constructing a three-stage radio and television news keyword and abstract combined extraction system by utilizing the trained multi-modal information fusion radio and television news keyword and abstract combined extraction model so as to perform keyword and abstract extraction on the input radio and television news text. According to the method, the keyword and the abstract are taken as a joint task, so that information sharing and feature fusion are facilitated, and the extraction effect of the keyword and the abstract can be improved in combination with the matts font image.
Description
technical field [0001] The invention relates to the technical field of news text content information extraction, in particular to a method for joint extraction of keywords and abstracts of radio and television news with multimodal information fusion. Background technique [0002] In the era of converged media in recent years, the explosive growth of video data has brought huge challenges to the reuse of media resources, and the real-time requirements for program cataloging and indexing have become higher accordingly. It is of great significance to use big data and artificial intelligence technology to automatically classify, identify, and index media content in multiple dimensions, realize the automatic extraction of content tags of media resources themselves, and improve cataloging quality and work efficiency. This is the supporting basis for the gradual transformation of media data management from traditional manual cataloging to automatic cataloging relying on an intellig...
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 Patent Type & Authority Applications(China)
IPC IPC(8): G06F16/33G06F16/335G06F40/126G06F40/289G06F40/30G06N3/04G06N3/08
CPCG06F16/3344G06F16/335G06F40/126G06F40/289G06F40/30G06N3/084G06N3/044
Inventor 杨瀚朱婷婷张阳林
Owner CHENGDU SOBEY DIGITAL TECH CO LTD