Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Semantic fuzzy search method based on sentence-level deep learning language model

A language model and deep learning technology, applied in semantic analysis, unstructured text data retrieval, special data processing applications, etc., can solve the problem that texts with different language structures cannot be matched, cannot accurately calculate semantic similarity, and cannot be efficiently It can improve the recall rate, fast operation speed, and convenient invocation.

Active Publication Date: 2020-04-10
深圳前海黑顿科技有限公司
View PDF7 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] 1. Most of the current text fuzzy search cannot solve the problem of semantic understanding of the text, which makes the search recall rate low, because if the problem of semantic understanding is not considered, many texts with the same meaning but different language structures cannot be matched;
[0005] 2. Most of the current fuzzy text searches are relatively inefficient. When searching for keywords or key sentences in relatively long texts, the efficiency is relatively low due to the use of violent methods to process the text;
[0006] 3. The current text fuzzy search cannot solve the problem of semantic deviation of keywords or key sentences caused by the context in the text when analyzing the semantic understanding of the text, which will reduce the recall rate of the search, and in the When analyzing semantic similarity, a relatively single similarity measurement index is used, which cannot accurately calculate the similarity between semantics, that is, it cannot efficiently measure the correlation between semantics

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
  • Semantic fuzzy search method based on sentence-level deep learning language model
  • Semantic fuzzy search method based on sentence-level deep learning language model
  • Semantic fuzzy search method based on sentence-level deep learning language model

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0038] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention.

[0039] refer to Figure 1-3 , the present invention also proposes a method for semantic fuzzy search based on a sentence-level deep learning language model, comprising the following steps:

[0040] S1. Build an application scenario. Given a long text S and a query sentence Q, it is necessary to query the string most relevant to Q in S;

[0041] S2. Build a language model library, train or directly call pre-trained sentence-level deep learning language model methods, such as: ELMo (Embeddings from Language Models), BERT (Bidirectional Encoder Representations from Transformers), etc., and uniformly adjust their operating mechanisms;

[0042] S3. Set the custom termina...

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 semantic fuzzy search method based on a sentence-level deep learning language model. According to the method, the fuzzy degree is high, a deep learning language model is introduced, the semantic problem is fully considered, statements with high semantic similarity with target statements can be retrieved, and the semantic similarity between the statements is efficiently judged in a layered semantic similarity calculation mode; the operation speed is high, vectorization processing is used for replacing a conventional text cyclic traversal processing mode, all semantic matching task units are ensured to be processed in parallel, and the search speed is greatly increased; the search recall ratio is high, and implication indexes are utilized, so that the robustness ofthe system to grammar interference is better, and the search recall ratio is effectively improved; the system is flexible, the mechanisms of semantic comprehension, fuzzy query, accurate information positioning and the like are fused, and then the whole algorithm module is subjected to interface encapsulation, so that convenience is brought to a user to call.

Description

technical field [0001] The present invention relates to the relevant field of deep learning language model and fuzzy search, in particular to a semantic fuzzy search method based on sentence-level deep learning language model. Background technique [0002] Language models are widely used. As early as the 1970s, statistical language models have been successfully applied. In 2003, some scholars proposed to apply the concept of word vector to neural network-based language models. However, This kind of model has many parameters, and the training is more complicated; in 2010, some scholars proposed to apply the Recurrent Neural Network (Recurrent Neural Network) to the language model, which opened the wide use of deep learning in the language model, and later A series of excellent language models have emerged, such as: ELMO (Embedding from Language Models) model, Transformer model, and BERT (Bidirectional Encoder Representations from Transformer) model. Text fuzzy search is appl...

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): G06F16/33G06F40/30
CPCG06F16/3344
Inventor 裴正奇段必超黄梓忱朱斌斌段朦丽于秋鑫
Owner 深圳前海黑顿科技有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products