The patent technology of this invention discloses a question-and-answer methodology based on diabetes knowledge mapping, which includes: a question-and-answer system based on target data, preset data formats, and types of entities, relationships and attributes; an entity identification of target data based on data formats; a knowledge extraction based on relationship type and attribute type, based on entity identification results, and obtained a target three-track group; storage of target three blocks into the target map database to complete the construction of the target knowledge map; pre-processing of input issues to determine the target entity and type of target relationship; and searching of target map databases based on target entities and target relationships to obtain answers to the target recommended outcome. The patent technology of this invention can effectively improve the accuracy of the entity identification distributed in the target data, improve the accuracy of the answers in the question and answer system and be widely applied in the field of knowledge mapping techniques. Domain. Domain。
Download all detailed technical information
[summary of technical realization steps]
A question-and-answer system based on diabetes mapping
This patent technology involves knowledge maps spectrum
In particular, a question-and-answer system based on diabetes mapping。
Technical presentation
There are currently fewer studies of the construction of knowledge maps in the field of diabetes, most of the data for the construction of the profiles are derived from web-based reptiles or related books, and more methods are available for processing the above-mentioned unstructured data, while studies of non-structured data processing also have difficulties in extracting entities with long durations that are linked to the distribution across sentences. The often constructed knowledge map does not accurately reflect the various entities and their relationships。
In addition, the search engine is often used for queries, although feedback from existing search engines usually includes web pages and even advertisements, especially in specialized areas, which do not allow accurate feedback to search for the corresponding points, while the question-and-answer system, as a new type of information retrieval technology, is able to return directly to the user for accurate answers, thus saving the user time to search for the required information from a large number of relevant web pages。
Therefore, the question of how to construct an accurate mapping of the knowledge of the entity and the relationship, and thus the search for accurate problem systems, is an urgent issue。
Technical realization thinking
With this in mind, this patent technology implementation example provides an efficient, question-and-answer system based on diabetes knowledge mapping
Technical protection points
Technical profile summary
1. A question-and-answer system based on diabetes mapping is structured along the following characteristics: based on target data, predefined data formats, and the type of entity, relationship type and attribute; physical identification of the targeted data according to the data format described; knowledge extraction based on the type of relationship described and the type of attribute described, based on the results identified by the entity in question, and receiving the three-tier group of objectives; storage of the three components of the stated objective in the target map database to complete the construction of the target knowledge map; pre-processing of input issues to determine the target entity and type of target relationship; and searching of the targeted map database according to the target entity described and the type of target relationship described, to obtain the answers recommended for the target results. 2. A question-and-answer system based on diabetes, as described in claim 1, is structured along the following lines: 3. A question-and-answer system based on diabetes mapping, as described in claim 1, is structured along the lines of target-based data, predefined data formats, and entity categories, relationship types and attributes, including: diakg, a chinese data set based on diabetes mapping, data formats for preset array structures, and 18 entity categories, 16 relationship types and attributes. 4. The characteristic of a question-and-answer system based on diabetes mapping, as described in claim 1, is the physical identification of the stated target data on the basis of the described data format, based on the specified entity categories, including: cross-wording of the stated target data on the basis of the three referenced data formats; and cross-wording of the said target data on the basis of the stated cross-word description
Lstm
Crf models for physical identification. 5. A question-and-answer system based on diabetes mapping, as described in claim 1, is structured along the following characteristics: the knowledge extraction based on the identification of the entity in question, the knowledge extraction based on the type of relationship in question and the type of attribute in question, and the target triform, including: the relationship extraction based on the identification of the entity in question, based on the type of relationship in question, and the relationship-oriented text-oriented extraction based on the bert model; and, based on ...
[property technical properties]
Technical researcher: hao tianyong, lan sheng, zhou ying ying
The application (patent) is made by the university of the south china teachers college
Type: invention
Country and city:
I'm the owner of this patent
