Large-scale power affair graph processing method based on expansion model

Abstract

The invention discloses a large-scale electric power affair graph processing method based on an expansion model, and the method comprises the steps: carrying out the statistics of all event expansion requests received in a preset time period through the expansion model, and generating a first expansion node set; node type labels are added to all the extended affair nodes according to the required argument set and the selectable argument set of each extended affair node; carrying out secondary sorting on the extended event nodes in each classification subset to obtain a second extended node set; determining a corresponding first affair node and a corresponding second affair node according to the necessary argument set and the selectable argument set of each piece of expansion event information; counting the first affair node and the second affair node obtained at the previous moment to generate a determined node set, and obtaining node positions of the first affair node and the second affair node corresponding to the extended affair node traversed at the current moment based on the determined node set, and respectively connecting the extended affair node with the first affair node and the second affair node.

Description

technical field [0001] The invention relates to the technical field of data processing, in particular to a large-scale power transaction graph processing method based on an extended model. Background technique [0002] The knowledge in the traditional knowledge graph is static, describing the relationship between entities and entities. These relationships are relatively definite static knowledge, while in the real human society, knowledge is dynamic, and knowledge itself will be affected by external conditions. change and distortion. The knowledge graph has limitations in application. The knowledge graph can only answer the question of what is what. The knowledge reasoning including inferring unknown knowledge based on known knowledge and verifying the correctness of known knowledge fundamentally does not jump out. The "static" property. On questions such as "what's wrong", "what will happen next", "why", "how to do", etc., the knowledge graph is a little weak. The event ...

AI Technical Summary

Problems solved by technology

The former has high accuracy, but the construction cost is high, and the scale is difficult to grow rapidly; the latter is what you see is what you get, the construction cost is low, the scale can be rapidly expanded, and a large amount of logic can be quickly excavated, but the disadvantage is that the accuracy is affected by many factors , the accuracy rate is lower than the former

[0004] No matter what type of event map is constructed, it is necessary to continuously update the event nodes and events in the event map in the subsequent use process. When dividing the nouns in the event, it can be divided into mandatory arguments and optional arguments. Optional argument, the mandatory argument is the inevitable subject to complete the whole event, so the entity node corresponding to the inevitable subject has a strong correlation with the corresponding event node, and the optional argument is the non-necessary subject to complete the entire event, so the The entity node corresponding to the non-necessary subject has a weak correlation with the corresponding event node. The argument can reflect the expansion relationship of the two new nodes in the previous event map. If the optional argument of the two new nodes The closer the elements and / or mandatory arguments are, the closer the previous event nodes of the graph connected by the two newly added event nodes will be. Currently, there is no technical solution that can , orderly and effectively update the map of affairs

Images