Knowledge mapping is a network of relationships that connect all kinds of information (heterogeneous information)。
Similar to what we used to write in our notes, it can also be called a knowledge map。
(2) role of knowledge mapping
The depth of the first application by knowledge mapping can be divided into two main categories:
The first is a generic knowledge map, with no particularly deep industrial knowledge and professional content, which generally addresses such issues as general knowledge and science. More commonly, there are recommendations in search engines, chat robots, etc。
The second is an industry knowledge map, popularly known as a professional version, customized on the basis of an in-depth study of a particular industry or subdivision area, which addresses, inter alia, professional issues in the current industry or subdivision. It has been widely applied in various vertical areas, such as finance, agriculture, electricians, health care, environmental protection and industrial manufacturing。
(3) how to build a knowledge map spectrum
Building a knowledge mapping system requires three steps: knowledge modelling, knowledge storage and knowledge application:
1. Knowledge modelling: first, it is clear whether the current business is suitable, needs to use the knowledge mapping and to define, organize, manage and model the corresponding knowledge mapping by using abstract information such as knowledge, attributes, linkages, etc. In the business. Convert data from different sources and structures into spectrograph data. There is also sometimes a need to integrate multiple sources, duplicate knowledge information, including integration computing, integration computing engines, manual operations, etc。
Knowledge storage: a sound knowledge storage programme based on the business landscape that is flexible, diverse and accessible. We have to face the option of storing the storage system, but because of the properties of the knowledge maps that we design, graphic databases, such as neo4j, can be preferred. However, the selection of which chart database would also depend on the volume of operations and efficiency requirements. If the amount of data is particularly large, neo4j is likely to fail to meet business needs, at which point it will be necessary to select distribution-support systems such as orientdb, janusgraph (formerly titan), or to store information in traditional databases through efficiency and redundancy, thereby reducing the amount of information contained in knowledge maps. Usually neo4j is enough。
5. Knowledge applications: providing analytical and application capabilities such as mapping retrieval, knowledge computing, mapping visualization, etc. For built knowledge maps. Sdk, which also provides various knowledge calculations, builds relevant applications using relevant technologies such as ai, big data, at the top of the chart database. It is also possible to exploit the implicit correlations in the knowledge map by using a library of algorithms attached to the knowledge map. In the example of neo4j, the algorithms that can be supported by neo4j-algorithms have similarity calculations (cosine similarity, jaccard similarity, pearson pilson similarity), minimum path algorithms, community discovery, centre calculation, etc. The results of the algorithm analysis are presented。
