CN115881106A - Intelligent form filling method, equipment and medium based on multi-scene semantic analysis - Google Patents

Abstract

The embodiment of the specification discloses an intelligent form filling method, equipment and a medium based on multi-scene semantic analysis, and relates to the technical field of computers, wherein the method comprises the following steps: acquiring voice document information of a document to be filled by a user, carrying out primary identification on the voice document information, and generating a voice document scene corresponding to the voice document information; performing secondary identification on the voice document information through a preset expert database corresponding to the voice document scene according to the voice document scene, and acquiring current form filling data corresponding to the voice document scene; determining appointed target information in the voice document information, and performing semantic extraction on the appointed target information to generate a current form filling abstract; and intelligently filling the bill to be filled through the current filling data and the current filling abstract. The method has the advantages that the semantics, namely the scene, the expert database bring out peripheral information, and the intelligent form filling method is automatically triggered, so that the filling error rate and the initial review return rate are reduced, the whole-person account-reporting training cost is reduced, the account-reporting efficiency is improved, and the complicated operation steps of a user are avoided.

Description

Intelligent form filling method, equipment and medium based on multi-scene semantic analysis

Technical Field

The present disclosure relates to the field of computer technologies, and in particular, to an intelligent form filling method, device, and medium based on multi-scenario semantic analysis.

Background

With the gradual popularization of mobile applications, cost control of mobile terminals has become a necessary trend of development, and this method is being adopted and used by more enterprises, and mobile terminal applications deployed in cloud terminals also enable small and medium-sized enterprises to participate in digital cost control. The traditional account-reporting system is complicated in form filling, under the condition of personalized management and control, the filling difference is large, the learning cost of a filling person is increased, the associated information is quickly matched according to consumption or filling items for form filling, and the simplified form filling process is a necessary course for promoting the mobile form filling to fall to the ground better. When the mobile terminal application deployed at the cloud end is used for filling the order, the step of filling the order is complicated, the user for filling the order needs to be trained, the learning cost of the user for filling the order is increased, and the accuracy of the order filling cannot be ensured.

Disclosure of Invention

One or more embodiments of the present specification provide an intelligent form filling method, device, and medium based on multi-scenario semantic analysis, which are used to solve the following technical problems: when the mobile terminal application deployed by the cloud is used for filling the order, the step of filling the order is complicated, the user who fills the order needs to be trained, the learning cost of the user who fills the order is increased, and the accuracy of the order filling cannot be ensured.

One or more embodiments of the present disclosure adopt the following technical solutions:

one or more embodiments of the present specification provide an intelligent form filling method based on multi-scenario semantic analysis, the method including: acquiring voice document information of a document to be filled by a user, carrying out primary identification on the voice document information, and generating a voice document scene corresponding to the voice document information; according to the voice document scene, performing secondary identification on the voice document information through a preset expert database corresponding to the voice document scene, and acquiring current form filling data corresponding to the voice document scene; determining appointed target information in the voice bill information, performing semantic extraction on the appointed target information, and generating a current form filling abstract; and intelligently filling the bill to be filled according to the current filling data and the current filling abstract.

Further, before performing secondary recognition on the voice document information through a preset expert database according to the voice document scene, the method further comprises the following steps: the method comprises the steps of obtaining a plurality of bill data and bill information corresponding to each bill data in advance, wherein the bill information comprises bill large-class information and preset project data; according to the bill large-class information in the bill information corresponding to each bill data, carrying out scene distinguishing on each bill data, and determining the bill initial scene corresponding to each bill data; performing secondary subdivision on the bill initial scene according to the preset project data, and determining a bill application scene corresponding to each bill data; and performing reverse optimization of semantic association through a principal component analysis method according to the document application scene to construct the preset enterprise library.

Further, according to the voice document scene, performing secondary recognition on the voice document information through a preset expert database, and acquiring current form filling data corresponding to the voice document scene, specifically including: identifying the voice document information to obtain document data information corresponding to the voice document information, wherein the document data information comprises document data time and document data types, and the document data types comprise any one or more of invoices, message records, consumption records and approval records; pulling a plurality of document source data corresponding to the document data information in a document data information corresponding system according to the document data information through the preset expert database; and matching the plurality of document source data according to the voice document scene through the preset expert database to obtain current filling data corresponding to the voice document scene.

Further, preliminarily identifying the voice document information, and generating a voice document scene corresponding to the voice document information, wherein the method specifically comprises the following steps: extracting key semantic information of the voice document information, wherein the key semantic information is used for representing an application scene of a document; identifying the key semantic information to generate scene semantics; and determining a voice document scene corresponding to the scene semantics according to the scene semantics and a preset first mapping relation, wherein the first mapping relation comprises a plurality of scene semantics and the voice document scene corresponding to each scene semantics.

Further, determining the specified destination information in the voice document information, performing semantic extraction on the specified destination information, and generating a current form filling abstract, specifically comprising: presetting a plurality of designated semantic fields, wherein the designated semantic fields are used for expressing purposes; splitting the voice document information to obtain a plurality of voice messages; calculating the similarity between the voice field corresponding to the voice information and the designated semantic field; determining a designated voice field with the similarity larger than a preset similarity threshold in a plurality of voice fields, and taking the designated voice field as the designated destination information; and extracting the semantics in the specified target information to generate a current form filling abstract.

Further, the intelligent form filling is performed on the to-be-filled document through the current form filling data and the current form filling abstract, and the method specifically comprises the following steps: filling the application items corresponding to the to-be-filled document according to the current filling abstract; and perfecting the bill source data corresponding to the bill to be filled according to the current filling data so as to realize intelligent filling of the bill to be filled.

Further, after the intelligent form filling is performed on the to-be-filled document through the current form filling data and the current form filling abstract, the method further includes: acquiring service associated data in the current form filling data; according to the business associated data, completing the associated information of the bill to be filled, wherein the completing the associated information of the bill to be filled specifically comprises the following steps: triggering an expression, perfecting a cost project and distributing information.

Further, according to the document application scenario, reverse optimization of semantic association is performed through a principal component analysis method, and the preset enterprise library is constructed, specifically including: performing reverse optimization of semantic association through a principal component analysis method according to the document application scene to generate an optimized semantic recognition result corresponding to the document application scene and a current scene recognition model corresponding to the document application scene; and determining the preset enterprise library according to the optimized semantic recognition result and the current scene recognition model.

One or more embodiments of the present specification provide an intelligent form filling device based on multi-scenario semantic analysis, including:

at least one processor; and the number of the first and second groups,

a memory communicatively coupled to the at least one processor; wherein,

the memory stores instructions executable by the at least one processor to enable the at least one processor to:

acquiring voice document information of a document to be filled by a user, performing primary identification on the voice document information, and generating a voice document scene corresponding to the voice document information; according to the voice document scene, performing secondary recognition on the voice document information through a preset expert database corresponding to the voice document scene, and acquiring current form filling data corresponding to the voice document scene; determining appointed target information in the voice document information, performing semantic extraction on the appointed target information, and generating a current form filling abstract; and intelligently filling the bill to be filled according to the current filling data and the current filling abstract.

One or more embodiments of the present specification provide a non-transitory computer storage medium storing computer-executable instructions configured to:

acquiring voice document information of a document to be filled by a user, carrying out primary identification on the voice document information, and generating a voice document scene corresponding to the voice document information; according to the voice document scene, performing secondary recognition on the voice document information through a preset expert database corresponding to the voice document scene, and acquiring current form filling data corresponding to the voice document scene; determining appointed target information in the voice document information, performing semantic extraction on the appointed target information, and generating a current form filling abstract; and intelligently filling the bill to be filled according to the current filling data and the current filling abstract.

The embodiment of the specification adopts at least one technical scheme which can achieve the following beneficial effects: according to the technical scheme, the voice document scene is determined through voice document information input by a user, document data are identified through the expert database according to the voice document scene, current form filling data corresponding to the scene are obtained, the semantics, namely the scene and the scene, are realized, the expert database brings peripheral information, and finally, the intelligent form filling method is automatically triggered, the filling error rate and the initial review return rate are reduced, the whole-person account reporting training cost is reduced, the account reporting efficiency is improved, the complicated operation steps of the user are avoided, and the user requirements are met.

Drawings

In order to more clearly illustrate the embodiments of the present specification or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, it is obvious that the drawings in the following description are only some embodiments described in the present specification, and for those skilled in the art, other drawings can be obtained according to the drawings without any creative effort. In the drawings:

fig. 1 is a schematic flowchart of an intelligent form filling method based on multi-scenario semantic analysis according to an embodiment of the present disclosure;

fig. 2 is a schematic structural diagram of an intelligent form filling device based on multi-scenario semantic analysis according to an embodiment of the present disclosure.

Detailed Description

In order to make those skilled in the art better understand the technical solutions in the present specification, the technical solutions in the embodiments of the present specification will be clearly and completely described below with reference to the drawings in the embodiments of the present specification, and it is obvious that the described embodiments are only a part of the embodiments of the present specification, and not all of the embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present specification without any creative effort shall fall within the protection scope of the present specification.

With the gradual popularization of mobile applications, cost control of mobile terminals has become a necessary trend of development, and this method is being adopted and used by more enterprises, and mobile terminal applications deployed in cloud terminals also enable small and medium-sized enterprises to participate in digital cost control. The traditional account-reporting system is complicated in form filling, under the condition of personalized management and control, the form filling difference is large, the learning cost of the form filling person is increased, the form filling is carried out by quickly matching the associated information according to consumption or filling items, and the simplified form filling process is a necessary repair course for promoting the mobile form filling to fall to the ground better. When the mobile terminal application deployed at the cloud end is used for filling the order, the step of filling the order is complicated, the user for filling the order needs to be trained, the learning cost of the user for filling the order is increased, and the accuracy of the order filling cannot be ensured.

The embodiment of the present specification provides an intelligent form filling method based on multi-scenario semantic analysis, and it should be noted that an executing subject in the embodiment of the present specification may be a server or any device with data processing capability, where form filling refers to filling a document. The embodiment of the specification is mainly used for a filling link of an ERP travelling expense control cloud product, and belongs to the field of distinguishing filling scenes (not traditional voice recognition environment scenes but filling document scenes) and carrying out scene matching and error correction on preliminary results of voice recognition; and performing semantic association matching according to the recognition error correction result, taking out related document contents according to a semantic association expert database, and simultaneously realizing document filling through an intelligent filling method. Fig. 1 is a schematic flowchart of an intelligent form filling method based on multi-scenario semantic analysis according to an embodiment of the present disclosure, and as shown in fig. 1, the method mainly includes the following steps:

step S101, acquiring voice document information of a document to be filled by a user, carrying out primary identification on the voice document information, and generating a voice document scene corresponding to the voice document information.

In an embodiment of the present specification, when a user needs to fill a bill, voice document information is input, for example, "i want to fill a business trip document for 11 months", the voice document information of the document to be filled by the user is obtained, the voice document information is preliminarily identified, and a voice document scene corresponding to the voice document information is determined, for example, a business trip document filling scene, a material document filling scene, a project expense document filling scene, or a wage expense document filling.

The voice document information is preliminarily identified, and a voice document scene corresponding to the voice document information is generated, and the method specifically comprises the following steps: extracting key semantic information of the voice document information, wherein the key semantic information is used for representing an application scene of a document; identifying the key semantic information to generate scene semantics; and determining a voice document scene corresponding to the scene semantics according to the scene semantics and a preset first mapping relation, wherein the first mapping relation comprises a plurality of scene semantics and the voice document scene corresponding to each scene semantics.

In an embodiment of the present specification, the key semantic information used for representing the document application scenario in the voice document information is extracted, and taking the travel document filling scenario as an example, the key semantic information used for representing the travel document filling application scenario may be "business document of 11 months", "travel document", "departure to a certain place", "go to a certain place", and the like. The key semantic information is recognized to obtain scene semantics, such as "business trip", "travel", "departure", "go", and the like. According to the obtained scene semantics, determining a voice document scene corresponding to the scene semantics in a preset first mapping relation, namely, according to the obtained scene semantics 'issue', obtaining a voice document scene 'travel document filling scene' corresponding to the document. In addition, it should be noted that the first mapping relationship is a pre-constructed corresponding relationship of the voice document scene for representing the scene semantics and each scene semantics, a plurality of scene semantics under each voice document scene may be listed in advance, one or more scene semantics corresponding to each voice document scene are obtained according to the voice document scene required by the enterprise in the history form filling process, and a corresponding relationship between the voice document scene and the scene semantics is established, where the corresponding relationship may be one-to-many or one-to-one. In addition, the voice document scene can be distinguished according to the account report unit besides the document type, wherein the document type is used for representing the voice document scene.

In an embodiment of the present specification, after the voice document scene is obtained, a document filling type corresponding to the voice document scene may be called according to the voice document scene, for example, after the travel document filling scene is obtained, a document type of the travel document is automatically selected, which may be understood as a template for the travel document filling. It should be noted that, the selection of the travel document filling template herein may also be selected by the user.

And S102, carrying out secondary recognition on the voice document information through a preset expert database corresponding to the voice document scene according to the voice document scene, and acquiring current form filling data corresponding to the voice document scene.

According to the voice document scene, before secondary recognition is carried out on the voice document information through a preset expert database, the method further comprises the following steps: the method comprises the steps of obtaining a plurality of bill data and bill information corresponding to each bill data in advance, wherein the bill information comprises bill large-class information and preset project data; according to the large bill information in the bill information corresponding to each bill data, carrying out scene distinguishing on each bill data, and determining the initial bill scene corresponding to each bill data; performing secondary subdivision on the bill initial scene according to the preset project data, and determining a bill application scene corresponding to each bill data; and performing reverse optimization of semantic association through a principal component analysis method according to the document application scene, and constructing the preset enterprise library.

According to the document application scene, reverse optimization of semantic association is carried out through a principal component analysis method, and the preset enterprise library is constructed, and the method specifically comprises the following steps: according to the receipt application scene, reverse optimization of semantic association is carried out through a principal component analysis method, and an optimized semantic recognition result corresponding to the receipt application scene and a current scene recognition model corresponding to the receipt application scene are generated; and determining the preset enterprise library according to the optimized semantic recognition result and the current scene recognition model.

In an actual application scenario, when filling out a bill, a plurality of peripheral data, such as a plurality of invoice data, are usually brought out if the business travel bill is required to be filled in within a certain time period, but if a user fills in the business travel bill, only business travel bills are required, workload of user screening is increased, and if the taken out invoice data are directly intelligently filled in, the bill filling is inaccurate.

In an embodiment of the present specification, before performing secondary recognition according to a voice document scene, an expert database needs to be established in advance, where the expert database may be understood as a rule base, the establishment of the expert database mainly depends on principal component analysis to summarize, and a plurality of document data and document information corresponding to each document data are obtained in advance, the document information includes document broad-class information and preset project data, and the project data is data preset by an enterprise user, for example, data required by a document required by a certain company is invoice data, but data required by other companies is consumption records. The bill initiator initiates different bills, the basic logics corresponding to the bills comprise bill large-class information and bill scene information, such as bill types, account reporting units and the like, different bill business scenes are distinguished and defined, and the business scenes mainly distinguish information and comprise time, bill types, personal type use and processing business preference and the like. The method improves the semantic analysis efficiency and the semantic result accuracy after voice recognition in the expert database establishing process, and simultaneously matches the expert database corresponding to the real use of the user as correctly as possible. After scenes are distinguished according to project categories of data deviation to large categories and manual intervention, main component data analysis under the scenes is carried out according to the scenes, so that a correct expert database can be searched in a relevant mode in the actual document filling process, the expert database can carry out analysis and prefabrication of scene data according to corresponding scenes, and invoices, message records and the like in corresponding time stages can be carried out according to semantics such as time, places and the like.

And distinguishing scenes of each bill data according to the bill large-class information in the bill information corresponding to the bill data, and determining an initial bill scene corresponding to each bill data. And carrying out secondary subdivision on the bill initial scene according to preset project data, and determining a bill application scene corresponding to each bill data. And after the scene is subdivided, reverse optimization of semantic association is carried out according to a principal component analysis method, and a semantic recognition result and a corresponding model of the expert database are confirmed, so that the voice document information input by a user is recognized through the corresponding model of the expert database subsequently according to the voice document scene, and analysis and prefabrication of data capable of being taken out of the scene are carried out according to the corresponding scene.

According to the voice document scene, secondary recognition is carried out on the voice document information through a preset expert library, current form filling data corresponding to the voice document scene are obtained, and the method specifically comprises the following steps: identifying the voice document information to obtain document data information corresponding to the voice document information, wherein the document data information comprises document data time and document data types, and the document data types comprise any one or more of invoices, message records, consumption records and approval records; pulling a plurality of document source data corresponding to the document data information in a system corresponding to the document data information according to the document data information through the preset expert database; and matching the plurality of document source data according to the voice document scene through the preset expert database to obtain current form filling data corresponding to the voice document scene.

In an embodiment of the present specification, identifying voice document information to obtain document data information corresponding to the voice document information, where the document data information includes document data time and document data type, and the document data type includes any one or more of invoice, message record, consumption record, and approval record, for example, the obtained document data information is an invoice in September, and pulling, by a preset expert database, a plurality of document source data corresponding to the document data information in a document data information corresponding system according to the document data information; that is to say, when the document data information is the September invoice, all invoices in September in the invoice folder are pulled through the expert database, namely, a plurality of document source data. And matching the plurality of document source data according to the voice document scene through a preset expert library to obtain current form filling data corresponding to the voice document scene. For example, after all invoices in September are obtained, because the voice document scene of the user is a travel bill filling scene, the travel invoices are obtained from all invoices in September according to the travel bill filling scene, and the travel invoices are current bill filling data.

In an embodiment of the present specification, identifying the document scene classification is to confirm a speech recognition scene according to the document category information and the document scene information, such as document type, account statement unit, and the like. The components of the speech recognition method are mainly realized by relying on a mature third-party recognition library. The training optimization of semantic recognition depends on recognition scenes to perform semantic recognition training to obtain a more appropriate semantic recognition result. Matching the expert database of the semantic recognition result, and confirming the corresponding expert database according to the association matching rule.

And step S103, determining appointed target information in the voice document information, performing semantic extraction on the appointed target information, and generating a current form filling abstract.

Determining appointed target information in the voice document information, performing semantic extraction on the appointed target information, and generating a current form filling abstract, wherein the method specifically comprises the following steps: presetting a plurality of designated semantic fields, wherein the designated semantic fields are used for expressing purposes; splitting the voice document information to obtain a plurality of voice messages; calculating the similarity between the voice field corresponding to the voice information and the designated semantic field; determining a designated voice field with the similarity larger than a preset similarity threshold in a plurality of voice fields, and taking the designated voice field as the designated destination information; and extracting the semantics in the specified target information to generate the current form filling abstract.

In an actual form filling scene, except for filling corresponding form filling data such as invoice and consumption record and other voucher information in a form, a form filling abstract is required to be set for a current form filling.

In one embodiment of the present specification, a plurality of specified semantic fields for presentation purposes, such as "for", "because", and the like, are set in advance. Splitting the voice document information, wherein a splitting rule at the position can be set according to requirements, the purpose is to split the voice document information into a plurality of fields to respectively compare appointed semantic fields, calculating the similarity between the voice field corresponding to the split voice information and the appointed semantic fields, selecting the appointed voice fields with the similarity larger than a preset similarity threshold, determining the similarity threshold at the position according to a similarity calculation mode, and setting the similarity calculation mode according to user requirements. And taking the obtained specified voice field as specified target information, extracting semantics in the specified target information, and generating a form filling abstract.

And step S104, intelligently filling the bill to be filled according to the current filling data and the current filling abstract.

Through the current form filling data and the current form filling abstract, the intelligent form filling is carried out on the to-be-filled document, and the method specifically comprises the following steps: filling the application items corresponding to the to-be-filled document according to the current filling abstract; and according to the current form filling data, perfecting the document source data corresponding to the document to be filled so as to realize intelligent form filling of the document to be filled.

In one embodiment of the present specification, the abstract is filled in according to the abstract of semantic identification or the semantic of inductive property, and meanwhile, the intelligent filling method is triggered to complete the document information, the current filling data, that is, the information of the invoice, the consumption record and the like, is completed into the document source data, and the current filling abstract is completed into the application item.

After the intelligent form filling is performed on the to-be-filled document through the current form filling data and the current form filling abstract, the method further comprises the following steps: acquiring service associated data in the current form filling data; according to the business associated data, the associated information of the to-be-filled document is perfected, wherein the perfecting of the associated information of the to-be-filled document specifically comprises: triggering an expression, perfecting a cost project and distributing information.

In an embodiment of the present specification, since an application scenario of a form filling is enterprise daily management, when a form is filled, a form filling content may exist as a branch item of a certain item or other business requirements, and in order to meet such a situation, business-related data in current form filling data needs to be obtained, where the business-related data may be business information or a name of the item. And according to the business associated data, perfecting the associated information of the bill to be filled, such as triggering an expression, perfecting a cost item, allocating information and the like.

The embodiment of the specification also provides a specific implementation method corresponding to the embodiment. Firstly, relevant precondition parameters need to be set for realizing voice recognition and semantic association scenes. Defining related variable information: billClass is identified by the large type of the document, and a unique identification of the document type is defined: billType, the definition system single unique identifier: peoples id, reimbursement unit: DEPT; confirming the entries of the matching expert database as formatted entries of the identified result. And, a third party speech recognition library is used as a running framework for speech recognition. And (4) using the scene as a distinguishing library of a third-party semantic analysis training library to perform training without the scene.

Secondly, realizing semantic corresponding relation of an expert database and the expert database, defining an N-dimensional vector, directly defining main table data, defining sub-tables by adding line number marks, and taking project structure data as a fixed component; the principal component is defined to occupy a minimum proportion of Components =0.95. Thereafter, a running framework is established that supports principal component analysis. Standardizing all extracted data to ensure that the data standard is available; centralizing the average value of each variable subtracted from each variable; calling a function to calculate a covariance matrix and an eigenvalue and an eigenvector thereof; and calling an interface to restore the original data set. And obtaining a new expert database subentry according to the extracted main components. Finally, the abstract and the triggering intelligent form filling method are realized, and sentences containing the abstract or the induction property semantics are extracted from the semantic analysis result. And distinguishing data main input parameters used as intelligent method calling according to a speech recognition scene, and using other data as auxiliary variables.

According to the technical scheme, the voice document scene is determined through voice document information input by a user, document data are identified through the expert database according to the voice document scene, current form filling data corresponding to the scene are obtained, the semantics, namely the scene and the scene, are realized, the expert database brings peripheral information, and finally, the intelligent form filling method is automatically triggered, the filling error rate and the initial review return rate are reduced, the whole-person account reporting training cost is reduced, the account reporting efficiency is improved, the complicated operation steps of the user are avoided, and the user requirements are met.

An embodiment of the present specification further provides an intelligent form filling device based on multi-scenario semantic analysis, as shown in fig. 2, the device includes: at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to cause the at least one processor to:

acquiring voice document information of a document to be filled by a user, carrying out primary identification on the voice document information, and generating a voice document scene corresponding to the voice document information; according to the voice document scene, performing secondary recognition on the voice document information through a preset expert database corresponding to the voice document scene, and acquiring current form filling data corresponding to the voice document scene; determining appointed target information in the voice document information, performing semantic extraction on the appointed target information, and generating a current form filling abstract; and intelligently filling the bill to be filled according to the current filling data and the current filling abstract.

Embodiments of the present description also provide a non-volatile computer storage medium storing computer-executable instructions configured to:

acquiring voice document information of a document to be filled by a user, carrying out primary identification on the voice document information, and generating a voice document scene corresponding to the voice document information; according to the voice document scene, performing secondary identification on the voice document information through a preset expert database corresponding to the voice document scene, and acquiring current form filling data corresponding to the voice document scene; determining appointed target information in the voice document information, performing semantic extraction on the appointed target information, and generating a current form filling abstract; and intelligently filling the bill to be filled according to the current filling data and the current filling abstract.

The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the apparatus, device, and non-volatile computer storage medium embodiments, since they are substantially similar to the method embodiments, the description is relatively simple, and reference may be made to the partial description of the method embodiments for relevant points.

The foregoing description has been directed to specific embodiments of this disclosure. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.

The devices and the media provided in the embodiments of the present description correspond to the methods one to one, and therefore, the devices and the media also have beneficial technical effects similar to the corresponding methods.

As will be appreciated by one skilled in the art, embodiments of the present description may be provided as a method, system, or computer program product. Accordingly, the description may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the description may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The description has been presented with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the description. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.

The memory may include forms of volatile memory in a computer readable medium, random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). Memory is an example of a computer-readable medium.

Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Disks (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium, which can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.

It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising a … …" does not exclude the presence of another identical element in a process, method, article, or apparatus that comprises the element.

The above description is merely one or more embodiments of the present disclosure and is not intended to limit the present disclosure. Various modifications and alterations to one or more embodiments of the present description will be apparent to those skilled in the art. Any modification, equivalent replacement, improvement or the like made within the spirit and principle of one or more embodiments of the present specification should be included in the scope of the claims of the present specification.

Claims (10)

1. An intelligent form filling method based on multi-scenario semantic analysis is characterized by comprising the following steps:

acquiring voice document information of a document to be filled by a user, carrying out primary identification on the voice document information, and generating a voice document scene corresponding to the voice document information;

according to the voice document scene, performing secondary recognition on the voice document information through a preset expert database corresponding to the voice document scene, and acquiring current form filling data corresponding to the voice document scene;

determining appointed target information in the voice document information, performing semantic extraction on the appointed target information, and generating a current form filling abstract;

and intelligently filling the bill to be filled according to the current filling data and the current filling abstract.

2. The intelligent form filling method based on multi-scenario semantic analysis according to claim 1, wherein before performing secondary recognition on the voice form information through a preset expert database according to the voice form scenario, the method further comprises:

the method comprises the steps of obtaining a plurality of bill data and bill information corresponding to each bill data in advance, wherein the bill information comprises bill large-class information and preset project data;

according to the bill large-class information in the bill information corresponding to each bill data, carrying out scene distinguishing on each bill data, and determining a bill initial scene corresponding to each bill data;

performing secondary subdivision on the bill initial scene according to the preset project data, and determining a bill application scene corresponding to each bill data;

and performing reverse optimization of semantic association through a principal component analysis method according to the document application scene to construct the preset enterprise library.

3. The intelligent form filling method based on multi-scenario semantic analysis according to claim 2, wherein secondary recognition is performed on the voice form information through a preset expert database according to the voice form scenario to obtain current form filling data corresponding to the voice form scenario, and the method specifically comprises the following steps:

identifying the voice document information to obtain document data information corresponding to the voice document information, wherein the document data information comprises document data time and document data types, and the document data types comprise any one or more of invoices, message records, consumption records and examination and approval records;

pulling a plurality of document source data corresponding to the document data information in a document data information corresponding system according to the document data information through the preset expert database;

and matching the plurality of document source data according to the voice document scene through the preset expert database to obtain current filling data corresponding to the voice document scene.

4. The intelligent form filling method based on multi-scenario semantic analysis according to claim 1, wherein the preliminary recognition is performed on the voice document information to generate a voice document scenario corresponding to the voice document information, specifically comprising:

extracting key semantic information of the voice document information, wherein the key semantic information is used for representing an application scene of a document;

identifying the key semantic information to generate scene semantics;

and determining a voice document scene corresponding to the scene semantics according to the scene semantics and a preset first mapping relation, wherein the first mapping relation comprises a plurality of scene semantics and a voice document scene corresponding to each scene semantics.

5. The intelligent form filling method based on multi-scenario semantic analysis according to claim 1, wherein specific destination information in the voice document information is determined, semantic extraction is performed on the specific destination information, and a current form filling abstract is generated, specifically comprising:

presetting a plurality of designated semantic fields, wherein the designated semantic fields are used for expressing purposes;

splitting the voice document information to obtain a plurality of voice messages;

calculating the similarity between the voice field corresponding to the voice information and the designated semantic field;

determining a designated voice field with the similarity larger than a preset similarity threshold in a plurality of voice fields, and taking the designated voice field as the designated target information;

and extracting the semantics in the specified target information to generate a current form filling abstract.

6. The intelligent form filling method based on multi-scenario semantic analysis according to claim 1, wherein the intelligent form filling is performed on the to-be-filled document through the current form filling data and the current form filling abstract, and specifically comprises:

filling the application items corresponding to the to-be-filled document according to the current filling abstract;

and perfecting the bill source data corresponding to the bill to be filled according to the current filling data so as to realize intelligent filling of the bill to be filled.

7. The intelligent form filling method based on multi-scenario semantic analysis according to claim 1, wherein after the intelligent form filling is performed on the to-be-filled document through the current form filling data and the current form filling abstract, the method further comprises:

acquiring service associated data in the current form filling data;

and perfecting the associated information of the to-be-filled document according to the business associated data, wherein the perfecting the associated information of the to-be-filled document specifically comprises: triggering an expression, perfecting a cost project and distributing information.

8. The intelligent form filling method based on multi-scenario semantic analysis according to claim 2, wherein reverse optimization of semantic association is performed through a principal component analysis method according to the form application scenario to construct the preset enterprise library, and specifically comprises:

performing reverse optimization of semantic association through a principal component analysis method according to the document application scene to generate an optimized semantic recognition result corresponding to the document application scene and a current scene recognition model corresponding to the document application scene;

and determining the preset enterprise library according to the optimized semantic recognition result and the current scene recognition model.

9. An intelligent form filling device based on multi-scene semantic analysis, the device comprising:

at least one processor; and the number of the first and second groups,

a memory communicatively coupled to the at least one processor; wherein,

the memory stores instructions executable by the at least one processor to cause the at least one processor to:

acquiring voice document information of a document to be filled by a user, carrying out primary identification on the voice document information, and generating a voice document scene corresponding to the voice document information;

according to the voice document scene, performing secondary identification on the voice document information through a preset expert database corresponding to the voice document scene, and acquiring current form filling data corresponding to the voice document scene;

determining appointed target information in the voice document information, performing semantic extraction on the appointed target information, and generating a current form filling abstract;

and intelligently filling the bill to be filled according to the current filling data and the current filling abstract.

10. A non-transitory computer storage medium storing computer-executable instructions configured to:

acquiring voice document information of a document to be filled by a user, carrying out primary identification on the voice document information, and generating a voice document scene corresponding to the voice document information;

according to the voice document scene, performing secondary identification on the voice document information through a preset expert database corresponding to the voice document scene, and acquiring current form filling data corresponding to the voice document scene;

determining appointed target information in the voice document information, performing semantic extraction on the appointed target information, and generating a current form filling abstract;

and intelligently filling the bill to be filled according to the current filling data and the current filling abstract.

Images