A method and electronic device for question answer extraction based on multi-layer perception
A multi-layer perception and problem-solving technology, applied in the fields of electrical digital data processing, instruments, and unstructured text data retrieval, etc., can solve problems such as complex reasoning process, separate processing of reasoning types, etc., and achieve improved effects
Active Publication Date: 2022-07-08
INST OF INFORMATION ENG CHINESE ACAD OF SCI
View PDF6 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
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
[0038] In order to make the above-mentioned features and advantages of the present invention more obvious and easy to understand, the following embodiments are given and described in detail with the accompanying drawings as follows.
[0039] The present invention classifies the inference categories in the HotpotQA data set. There are two main categories, bridging entity classes and comparison classes. If the question is a bridged entity class reasoning type, the model will process the answer prediction process into two subtasks stacked in two levels. The bridging entity representation and question and context content to lock the final answer; if the question is a comparative reasoning type, the model will process the answer prediction process as a subtask of two layers of hierarchical stacking, the first layer is to find through two search modules to find For the two related entities in the model, the second layer uses a comparison module to compare the entity representations...
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 present invention provides a method for extracting questions and answers based on multi-layer perception. Representation Q interacts with context representation P to obtain the document representation u related to the document and the document representation h that fuses the problem information; perform multi-layer perceptual classification on the problem representation u to obtain the reasoning type of the question, and according to the reasoning type, question representation u, document representation represents h and the subproblem c generated by representing Q t , obtain the attention distribution of the answer of the question in the target document, where t is the number of times of generating sub-questions; according to the attention distribution of the answer, the answer prediction result of the question is obtained. The invention progressively answers questions in the form of sub-question splitting, introduces a reasoning category classifier to control splitting, shares answers to questions, and improves the effect of reasoning reading comprehension.
Description
technical field [0001] The invention belongs to the field of natural language processing, and in particular relates to a question answer extraction method and electronic device based on multi-layer perception. Background technique [0002] Inferential reading comprehension is given a user multiple related documents for a question, and finding the answer to the question and the relevant evidence sentences from the documents. Reasoning about reading comprehension problems requires a model to combine the problem, reason about the semantic meaning of the text, and find relevant evidence sentences and final answers to the problem. The reasoning reading model can be divided into three categories as a whole. One is the memory network method, which simulates the reasoning process by iteratively updating the reasoning state; the other is the method based on the graph neural network, which performs inference through the update of the graph neural network; there are other methods base...
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 Patents(China)
IPC IPC(8): G06F16/332G06F16/36G06N3/04
CPCG06F16/3329G06F16/36G06N3/049G06N3/044
Inventor 林政付鹏刘欢王伟平孟丹
Owner INST OF INFORMATION ENG CHINESE ACAD OF SCI