Search

CN-121981099-A - AI dialogue form filling method and system based on interactive card

CN121981099ACN 121981099 ACN121981099 ACN 121981099ACN-121981099-A

Abstract

The application discloses an AI dialogue form filling method and system based on an interactive card. The method comprises the steps of obtaining natural language information input by a user, generating a structured interaction card according to the natural language information input by the user, wherein the structured interaction card comprises a form type identification area, a filled item abstract area and an item prompt area to be supplemented, obtaining interaction modes and feedback data used by the user after the structured interaction card is obtained, and updating the structured interaction card according to the interaction modes and the feedback data. In the AI dialogue flow, an interactive card is introduced as the medium of the structured dialogue. The card is not the final form, but a dynamic, operational task state visualization and interaction center. The method solves the problems of structural disjoint and semantic gap in the prior art through the primary identification, visual confirmation, bidirectional interaction and finally executed steady closed loop.

Inventors

  • ZHOU QIANG

Assignees

  • 成都冠唐科技有限公司

Dates

Publication Date
20260505
Application Date
20260126

Claims (10)

  1. 1. An AI dialogue form filling method based on an interactive card, which is characterized by comprising the following steps: acquiring natural language information input by a user; generating a structured interaction card according to the natural language information input by the user, wherein the structured interaction card comprises a form type identification area, a filled item abstract area and a to-be-supplemented item prompt area; The method comprises the steps of obtaining an interaction mode and feedback data used by a user after obtaining a structured interaction card; And updating the structured interaction card according to the interaction mode and the feedback data.
  2. 2. The AI dialog form filling method based on an interactive card of claim 1, the AI dialogue form filling method based on the interactive card is characterized by further comprising the following steps: Judging whether the updated structured interactive card meets the submitting condition, if so, then Generating final interaction information and feeding back to a user; and responding to the submitting instruction of the user, and sending the final interaction information to a back-end service system.
  3. 3. The interactive card-based AI dialog form filling method of claim 1 wherein the interactive mode includes a form replenishment mode; when the interaction mode is a form supplementing mode, the updating the structured interaction card according to the interaction mode and the feedback data comprises the following steps: identifying supplementary information input by a user in a prompt area of a to-be-supplemented item; The structured interaction card is updated with the supplemental information.
  4. 4. The interactive card-based AI dialog form filling method of claim 3 wherein the interaction pattern includes a complete filling pattern; when the interaction mode is a complete filling mode, the updating the structured interaction card according to the interaction mode and the feedback data comprises: Generating a structured interaction form; Identifying supplementary information input by a user in the structured interactive form; The structured interaction card is updated with the supplemental information.
  5. 5. The interactive card-based AI dialog form filling method of claim 4 wherein the interaction pattern includes a voice interaction pattern; when the interaction mode is a voice interaction mode, the updating the structured interaction card according to the interaction mode and the feedback data includes: Identifying voice information of a user, thereby obtaining identification information; The structured interaction card is updated with the identification information.
  6. 6. The interactive card-based AI dialog form filling method of claim 5 wherein the interactive mode includes a business type intelligent correction mode; When the interaction mode is a business type intelligent correction mode, the updating the structured interaction card according to the interaction mode and the feedback data comprises the following steps: Identifying triggering operation of a user on a form type identification area in the structured interactive card; Invoking an alternative service type list associated with the current service scenario; displaying the alternative service type list to the user in a hierarchical option form, and attaching core field description corresponding to each service type; receiving a target service type selected by a user, and performing cross-type field mapping and adaptation according to the target service type to obtain a synchronization result of compatible field filled data of an adaptation new service type, a prompt to be supplemented of an incompatible field and historical effective feedback data; Pulling a structured form template corresponding to the target service type, carrying out validity check and format conversion on the filled compatible field data, and generating a to-be-supplemented item with priority ordering on the incompatible field; Based on the compatible field data after verification and conversion, the new to-be-supplemented item and the synchronous historical effective data, the form type identification area, the filled item abstract area and the to-be-supplemented item prompt area of the structured interactive card are updated and sent to a user.
  7. 7. The interactive card-based AI dialog form filling method of claim 6, wherein the generating a structured interactive card from the user-entered natural language information comprises: Acquiring an enterprise business scene keyword cluster library, an enterprise business process dependency rule library and a field attribute rule library; Generating a four-dimensional semantic unit set, an anchoring service type and a basic word sequence according to the natural language information and the enterprise service scene keyword cluster library; obtaining a structured form template, a form template field set and a field attribute detail corresponding to the anchoring service type according to the anchoring service type; Generating a field association network with a dependency priority and a three-dimensional matching rule set of each field according to the form template field set, the field attribute detail and the enterprise business process dependency rule base; generating a semantic unit-field matching table, a field candidate value set and an unmatched character-to-be-filled candidate filling value set of the unique mapping according to the four-dimensional semantic unit set, the field association network, the three-dimensional matching rule set and the field attribute detail; Performing fuzzy information scene derivation according to the field candidate value set and the unmatched necessary-filling field candidate filling value set, so as to obtain an accurate field filling value set and a field set to be supplemented; and generating the structured interaction card according to the accurate field filling value set and the field set to be supplemented.
  8. 8. The interactive card-based AI dialog form filling method of claim 7, wherein generating the four-dimensional semantic unit set, the anchor business type, the base word sequence from the natural language information and the business scenario keyword cluster library comprises: performing word segmentation processing on natural language information by adopting a word segmentation algorithm based on a bidirectional maximum matching method, so as to obtain an initial word sequence; performing invalid word filtering on the initial word sequence through a preset business field stop word list, so as to obtain an effective word sequence; part-of-speech tagging is carried out on the effective word sequence through a conditional random field model, so that a structured basic word sequence is obtained; Traversing the structured basic word sequence, screening words marked as verbs and having occurrence frequency more than or equal to 10 times in an enterprise business scene keyword cluster library, and taking the words as core predicates, thereby obtaining a core predicate set; Identifying dominant components and modification components of the core predicate by syntactic dependency analysis by taking the core predicate as a center, and splitting semantic units according to an action-entity-attribute-limiting condition four-dimensional structure, so as to obtain a structured four-dimensional semantic unit set; acquiring each candidate type of keyword cluster according to an enterprise business scene keyword cluster library; Respectively calculating the intra-cluster overlap ratio value and the intra-cluster word coverage ratio value of each candidate service type; and acquiring an anchor scene according to the intra-cluster overlap ratio value and the intra-cluster word coverage ratio value of each candidate service type.
  9. 9. The interactive card-based AI dialog form filling method of claim 8, wherein the obtaining an anchor scene from the intra-cluster overlap ratio value, the intra-cluster word coverage value for each candidate traffic type comprises: Calculating the average value of the intra-cluster overlap ratio value and the intra-cluster word coverage ratio value of each candidate type, sorting according to the descending order of the average value, selecting a scene cluster corresponding to the candidate service type with the highest average value as an anchoring scene, and determining the service type associated with the scene cluster as the anchoring service type.
  10. 10. An AI dialogue form filling system based on interactive card, characterized in that, the AI dialogue form filling system based on the interactive card comprises: the natural language information acquisition module is used for acquiring natural language information input by a user; The structured interactive card acquisition module is used for generating a structured interactive card according to the natural language information input by the user, and the structured interactive card comprises a form type identification area, a filled item abstract area and a to-be-supplemented item prompt area; The feedback information acquisition module is used for acquiring an interaction mode and feedback data used by a user after the user acquires the structured interaction card; and the updating module is used for updating the structured interaction card according to the interaction mode and the feedback data.

Description

AI dialogue form filling method and system based on interactive card Technical Field The application relates to the technical field of artificial intelligence interaction, in particular to an AI dialogue form filling method and system based on an interactive card. Background With the development of artificial intelligence technology, intelligent voice assistants and chat robots have been widely used in various business handling scenarios, such as ticket booking, application, purchasing, etc. These services typically require calling the software API of the backend system and filling in electronic forms that are complex in structure. The prior art generally employs two mainstream interaction paradigms: 1. Pure natural language dialogue mode-the user provides information through natural language once or through multiple rounds of dialogue, AI systems attempt to understand user intent and call APIs directly within an open, unrestricted semantic space. However, this mode has the fundamental disadvantage that natural language is ubiquitous in information that is incomplete, ambiguous and ambiguous. More importantly, the intent recognition of the AI is disjoint from the specific form structure that needs to be filled in later. For example, when the "city" field in the form must be selected from a preset list, the user says "i want to go to that city with terracotta soldiers and horses", AI may not be able to map accurately to "western security city of shanxi province" among numerous business options. Multiple rounds of inquiry can complement information, but the process is tedious, the intention is easy to drift, and the task failure rate and the user frustration are high. 2. And directly jumping to a form mode, namely completely giving up a dialogue interface after AI identifies the user intention, and directly jumping to a complete and static traditional form page to enable the user to fill in manually. This way, the consistency and nature of the conversation is roughly broken, and the user needs to suddenly switch from a free, context-dependent conversation mode to a strange, strictly structured form filling mode, experiencing severe cracking. It solves the accuracy problem of API calls, but completely sacrifices the fluency and efficiency of dialog interactions. The core problem of the prior art is the intent to identify "structural disjoint" between and accurate execution, and the "semantic gap" between unstructured in natural language and structured data required by the API. Neither "pure dialog" nor "jump form" can create a smooth, efficient transition bridge between the two. Disclosure of Invention The invention aims to provide an AI dialogue form filling method based on an interactive card, which at least solves one technical problem. In one aspect of the present invention, there is provided an interactive card-based AI conversation form filling method, including: acquiring natural language information input by a user; generating a structured interaction card according to the natural language information input by the user, wherein the structured interaction card comprises a form type identification area, a filled item abstract area and a to-be-supplemented item prompt area; The method comprises the steps of obtaining an interaction mode and feedback data used by a user after obtaining a structured interaction card; And updating the structured interaction card according to the interaction mode and the feedback data. Optionally, the AI dialogue form filling method based on the interactive card further comprises: Judging whether the updated structured interactive card meets the submitting condition, if so, then Generating final interaction information and feeding back to a user; and responding to the submitting instruction of the user, and sending the final interaction information to a back-end service system. Optionally, the interaction mode includes a form replenishment mode; when the interaction mode is a form supplementing mode, the updating the structured interaction card according to the interaction mode and the feedback data comprises the following steps: identifying supplementary information input by a user in a prompt area of a to-be-supplemented item; The structured interaction card is updated with the supplemental information. Optionally, the interaction pattern includes a complete filling pattern; when the interaction mode is a complete filling mode, the updating the structured interaction card according to the interaction mode and the feedback data comprises: Generating a structured interaction form; Identifying supplementary information input by a user in the structured interactive form; The structured interaction card is updated with the supplemental information. Optionally, the interaction mode includes a voice interaction mode; when the interaction mode is a voice interaction mode, the updating the structured interaction card according to the interaction mode and the feedback data includes: