CN116610722A - Algorithm recommendation method based on knowledge graph in data mining - Google Patents

Abstract

The application discloses an algorithm recommendation method based on a knowledge graph in data mining, which comprises the following steps: (1) Constructing an algorithm domain knowledge base, implementing the algorithm domain knowledge base in three directions, preprocessing the algorithm domain knowledge base, and optimizing the strategy base by the algorithm; (2) constructing a knowledge tag in the data field; (3) establishing a data domain and algorithm domain association knowledge graph; (4) constructing a knowledge extraction algorithm support; (5) Constructing a fact experience library, and giving a result by a violation recommendation algorithm; (6) constructing a data access module; (7) And a knowledge acquisition module is constructed, so that the fact knowledge is convenient to maintain, an evaluation interface is recommended, information is returned, and the given fact is identified. According to the application, the decision engine supported by the knowledge graph automatically recommends, and a plurality of algorithms and optimal algorithms, so that the threshold selected by the data mining algorithm is greatly reduced, meanwhile, the algorithm matching degree risk caused by human supervisor reasons can be avoided, and the success rate of data mining is increased.

Description

Algorithm recommendation method based on knowledge graph in data mining

technical field

The invention relates to the technical field of data mining, in particular to an algorithm recommendation method based on a knowledge map in data mining.

Background technique

In the field of data analysis, the algorithm selection and parameter optimization of data analysis usually requires professional data scientists, algorithm engineers, and professionals in the field of data correspondence. Only by working together can we find a highly matching algorithm and corresponding optimized parameters.

At present, the usual data mining projects in the industry generally invite some highly educated and very good data thinking and some data analysis practitioners with data processing experience on the construction site, and match with algorithm engineers who understand the corresponding field, combined with senior personnel in the industry, Form a project team systematically. From multiple perspectives such as theoretical derivation and practical experience, multiple rounds of attempts were made to find one or several algorithmic solutions that improve matching.

Contents of the invention

The present invention provides a method for recommending algorithms based on knowledge graphs in data mining. The purpose is to use knowledge graphs, etc., through knowledge extraction and other methods, and the system automatically provides algorithms to match according to the definition of relevant domains.

The application is achieved through the following technical solutions:

The algorithm recommendation method based on knowledge map in data mining includes the following steps:

(1) Construct an algorithmic domain knowledge base, which is generally implemented in three directions;

(2) Build knowledge labels in the data field, including business knowledge labels;

(3) Establish a knowledge graph associated with the data domain and the algorithm domain;

(4) Construct knowledge extraction algorithm support, and design it as an open, priority graph model;

(5) Build a factual experience database, and recommend cases that violate the recommendation algorithm to give results but are better in line with reality as established facts;

(6) Build a data access module, support manual labeling of data domains and automatic statistics of data statistical features, and be able to do algorithm matching trials, and need to enter relevant instances of algorithm domains;

(7) Build a knowledge acquisition module to facilitate the maintenance of factual knowledge, as well as recommend and evaluate interfaces, return information, and identify established facts.

As a preferred embodiment, the step (1) is generally constructed according to three directions, including a preprocessing method knowledge base, an algorithm knowledge base, and an algorithm optimization strategy base.

As a preferred embodiment, the business knowledge tags included in the step (2) include semiconductor 8-inch cp map, panel Array factory; statistical domain tags, such as data range, data covariance, variance, data distribution characteristics; result tags, Such as classification requirements, regression requirements, optimal solution requirements.

As a preferred embodiment, the priority graph model in the step (4) includes Embeddig-based, path-based methods, and combined methods; it also supports empirical extraction of deep learning.

Beneficial effect:

The algorithm recommendation method based on the knowledge map in the data mining of the present invention adopts an automatic recommendation algorithm and a parameter optimization scheme to introduce the data to be mined, and the system automatically uses the decision engine supported by the knowledge map to automatically Recommendations, several algorithms and optimal algorithms greatly reduce the threshold for data mining algorithm selection, and at the same time avoid the risk of algorithm matching caused by human supervisors, and increase the success rate of data mining.

Description of drawings

Fig. 1 is a business flow chart of the algorithm recommendation method based on knowledge graph in data mining of the present invention.

Detailed ways

The embodiments of the present invention are described in detail below in conjunction with the accompanying drawings: this embodiment is implemented on the premise of the technical solution of the present invention, and detailed implementation methods and specific operating procedures are provided, but the protection scope of the present invention is not limited to the following the described embodiment.

As shown in Figure 1, the algorithm recommendation method based on knowledge graph in data mining includes the following steps:

(1) Construct the algorithm domain knowledge base, which is generally implemented in three directions, the preprocessing method knowledge base, the algorithm knowledge base, and the algorithm optimization strategy base;

(2) Build knowledge labels in the data field, including business knowledge labels, such as semiconductor 8-inch cp map, panel Array factory, etc.; statistical domain labels, such as data range, data covariance, variance, data distribution characteristics, etc.; result labels, such as Classification requirements, regression requirements, optimal solution requirements;

(3) Establish a knowledge graph associated with the data domain and the algorithm domain;

(4) Construct knowledge extraction algorithm support, and design it as an open type; priority graph model, such as based on Embeddig, path-based method, combined method; also supports empirical extraction of deep learning;

(5) Build a factual experience database, and recommend cases that violate the recommendation algorithm to give results but are better in line with reality as established facts;

(6) Build a data access module, support manual labeling of data domains and automatic statistics of data statistical features, and be able to do algorithm matching trials, and need to enter relevant instances of algorithm domains;

(7) Build a knowledge acquisition module to facilitate the maintenance of factual knowledge, as well as recommend and evaluate interfaces, return information, and identify established facts.

The basic principles and main features of the present invention and the advantages of the present invention have been shown and described above. Those skilled in the industry should understand that the present invention is not limited by the above-mentioned embodiments. What are described in the above-mentioned embodiments and the description only illustrate the principle of the present invention. Without departing from the spirit and scope of the present invention, the present invention will also have Variations and improvements are possible, which fall within the scope of the claimed invention. The protection scope of the present invention is defined by the appended claims and their equivalents.

Claims (4)

1. The algorithm recommendation method based on knowledge map in data mining, is characterized in that, comprises the following steps: (1) Construct an algorithmic domain knowledge base, which is generally implemented in three directions; (2) Build knowledge labels in the data field, including business knowledge labels; (3) Establish a knowledge graph associated with the data domain and the algorithm domain; (4) Construct knowledge extraction algorithm support, and design it as an open, priority graph model; (5) Build a factual experience database, and recommend cases that violate the recommendation algorithm to give results but are better in line with reality as established facts; (6) Build a data access module, support manual labeling of data domains and automatic statistics of data statistical features, and be able to do algorithm matching trials, and need to enter relevant instances of algorithm domains; (7) Build a knowledge acquisition module to facilitate the maintenance of factual knowledge, as well as recommend and evaluate interfaces, return information, and identify established facts. 2. the algorithm recommendation method based on knowledge map in the data mining according to claim 1, it is characterized in that, in the described step (1), generally construct according to three directions, comprise preprocessing method knowledge base, algorithm knowledge base, Algorithm optimization strategy library. 3. the algorithm recommendation method based on knowledge map in the data mining according to claim 1, is characterized in that, the business knowledge label that comprises in the described step (2) comprises semiconductor 8 cun cp map, panel Array factory; Statistic domain Labels, such as data range, data covariance, variance, and data distribution characteristics; result labels, such as classification requirements, regression requirements, and optimal solution requirements. 4. The algorithm recommendation based on the knowledge map in the data mining according to claim 1 Method, it is characterized in that, in described step (4), prior graph model comprises based on Embeddig, Path-based methods, combined methods; also support empirical extraction for deep learning.