Meteorological service interaction method and system combined with pre-training model and template generation

Abstract

The invention discloses a meteorological service interaction method and system generated by combining a pre-training model and a template, and the method comprises the steps: generating a training data set through the template in combination with data crawled by a network, and then recognizing a meteorological service entity in a user question through the pre-training model; based on the text similarity of word vectors, linking the identified entities to a weather map, determining a service scene in a user question, further querying a suggestion group related to the scene in a scene-oriented weather service database, and according to the semantic similarity between the weather suggestions and the user question, selecting proper weather suggestions and filling the weather suggestions back to a template, and generating a meteorological service statement and returning to the user. The meteorological service interaction system provided by the invention can provide personalized scene services for users and generate flexible and diversified service statements.

Description

technical field [0001] The invention belongs to the technical field of artificial intelligence, and in particular relates to a weather service interactive system generated by combining a pre-training model and a template. Background technique [0002] With the development of the mobile Internet, people can conveniently obtain various weather-related services from various websites and applications, and then plan various arrangements such as travel and home, which greatly improves people's quality of life. However, the current meteorological service can only provide users with simple and general service content, and cannot meet the user's refined scene service needs, and cannot answer the user's independent questions. On the other hand, with the continuous development of people's material and spiritual life, new weather service scenarios emerge in endlessly. However, the current weather service content relies on manual editing by experts. Once a new weather service scenario a...

AI Technical Summary

Problems solved by technology

However, in the weather service, the semantic correlation between the user's question and the answer that the system needs to provide is difficult to reflect, such as reminding the interacting user to bring an umbrella and collect clothes in time on rainy days, etc.

In addition, the grammar of such models often has some minor errors (such as easy to generate duplicate content, keep warm in rainy days)

Images