CN112115252A

CN112115252A - Intelligent auxiliary writing processing method and device, electronic equipment and storage medium

Info

Publication number: CN112115252A
Application number: CN202010871570.3A
Authority: CN
Inventors: 罗彤; 张学彬; 闻飞祥; 王雨农
Original assignee: Beijing Ronghui Jinxin Information Technology Co ltd
Current assignee: Beijing Ronghui Jinxin Information Technology Co ltd
Priority date: 2020-08-26
Filing date: 2020-08-26
Publication date: 2020-12-22
Anticipated expiration: 2040-08-26
Also published as: CN112115252B

Abstract

The embodiment of the invention discloses an intelligent auxiliary writing processing method, a device, electronic equipment and a storage medium, wherein the embodiment of the invention identifies the writing content of auxiliary data needing to be quickly inquired, acquires the inquiry intention identification result of the writing content of the auxiliary data needing to be quickly inquired, further inquires a knowledge graph according to the inquiry intention type identification result and the entity dependence relationship identification result, acquires the auxiliary data corresponding to the writing content and needing to be quickly inquired, and displays the auxiliary data corresponding to the writing content and needing to be quickly inquired in a specified area of a current writing page, so that the embodiment of the invention can provide intelligent writing assistance in the writing process of a user, analyzes the inquiry intention of the user on the writing auxiliary data from the writing content, further acquires the corresponding writing auxiliary data from a related knowledge graph according to the inquiry intention, therefore, the writing efficiency can be effectively improved, and intelligent writing becomes possible.

Description

Intelligent auxiliary writing processing method and device, electronic equipment and storage medium

Technical Field

The invention relates to the technical field of artificial intelligence, in particular to an intelligent auxiliary writing processing method and device, electronic equipment and a storage medium.

Background

Artificial Intelligence (AI), a new technical science to study and develop theories, methods, techniques and application systems for simulating, extending and expanding human Intelligence. Artificial intelligence is a branch of computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can react in a manner similar to human intelligence, a field of research that includes robotics, language recognition, image recognition, natural language processing, and expert systems, among others.

The traditional writing method mainly depends on manual mode to inquire and retrieve data and knowledge, and repeatedly draws and organizes data such as tables, pictures and the like, and the process is often complicated and inefficient. With the increasing development of deep learning, high-quality writing scenes (macro economy, macro strategy, industry, analysis reports and weekly reports of companies, company financial and newspaper comments, government reports, academic reports and the like) depend on a large amount of data and knowledge, and the whole process of title, material collection, writing, drawing, auditing and publishing is realized, so that the production efficiency of articles and reports is low, the quality can be guaranteed, and a lot of people are difficult to get rid of heavy labor.

Disclosure of Invention

Because the existing methods have the above problems, embodiments of the present invention provide an intelligent auxiliary writing processing method and apparatus, an electronic device, and a storage medium.

In a first aspect, an embodiment of the present invention provides an intelligent auxiliary writing processing method, including:

receiving writing content which is triggered and sent by a user in the writing process of the current writing page in a preset triggering mode and needs to quickly inquire auxiliary data;

performing query intention recognition on the writing content of the auxiliary data needing to be quickly queried, and acquiring a query intention recognition result of the writing content of the auxiliary data needing to be quickly queried; wherein the query intention identification result comprises a query intention category identification result and an entity dependency relationship identification result; the entity dependency relationship identification result is obtained by depending on the entity identification result and the dependency relationship among the entities;

inquiring a knowledge graph according to the inquiry intention type identification result and the entity dependence relationship identification result, and acquiring auxiliary data which corresponds to the writing content and needs to be inquired quickly;

and displaying the auxiliary data which is corresponding to the writing content and needs to be quickly inquired in a specified area of the current writing page so as to be selected and used by a user aiming at the auxiliary data in the writing process.

Further, performing query intention recognition on the writing content of the auxiliary data needing quick query, and acquiring a query intention recognition result of the writing content of the auxiliary data needing quick query, including:

performing query intention type recognition on the writing content of the auxiliary data needing to be quickly queried to obtain a query intention type recognition result;

performing entity identification and entity dependency relationship identification on the writing content of the auxiliary data needing to be quickly inquired, and acquiring an entity identification result and an entity dependency relationship identification result; wherein the entity recognition result at least comprises: an object entity, a time entity, and an index entity; the entity dependency relationship identification result at least comprises: a combination constraint between the object entity, the time entity and the index entity.

Further, performing query intention type recognition on the written content needing the quick query auxiliary data, and acquiring a query intention type recognition result, including:

inputting the writing content of the auxiliary data needing to be quickly inquired into a preset intention type identification model to obtain an inquiry intention type identification result;

the preset intention type identification model is obtained by taking writing content training samples of all known query intention types as sample input data, taking query intention types corresponding to the writing content of all known query intention types as sample output data and carrying out model training based on a machine learning algorithm; and the query intention type identification result is used for representing a field branch corresponding to the writing content of the auxiliary data needing quick query.

Further, performing entity identification and entity dependency relationship identification on the written content of the auxiliary data needing quick query, and acquiring an entity identification result and an entity dependency relationship identification result, including:

and performing sentence segmentation, word segmentation, part of speech tagging and dependency syntax analysis on the written content needing to quickly query the auxiliary data to obtain an entity identification result and an entity dependency relationship identification result.

Further, querying a knowledge graph according to the query intention type identification result and the entity dependency relationship identification result, and acquiring auxiliary data corresponding to the written content and needing quick query, including:

determining a knowledge graph of a corresponding domain branch according to the query intention type identification result;

acquiring auxiliary data which corresponds to the written content and needs to be quickly inquired from the knowledge graph of the corresponding field branch according to the entity dependency relationship identification result;

the knowledge graph of the corresponding domain branch stores knowledge and time sequence data of the corresponding domain branch, and the knowledge and time sequence data comprise each index information of each object in each time in the corresponding domain branch.

determining an extended entity dependency relationship identification result corresponding to the entity dependency relationship identification result according to the entity dependency relationship identification result; the object entity in the extended entity dependency relationship identification result and the object entity in the entity dependency relationship identification result have a class relationship or an auction relationship; the time entity in the extended entity dependency relationship identification result and the time entity in the entity dependency relationship identification result have an adjacent relationship or a periodic relationship; the index entity in the extended entity dependency relationship identification result and the index entity in the entity dependency relationship identification result have a similar relationship or an opposite relationship;

acquiring accurate auxiliary data which corresponds to the written content and needs to be quickly inquired from the knowledge graph of the corresponding field branch according to the entity dependency relationship identification result; acquiring related auxiliary data which corresponds to the writing content and needs to be quickly inquired from the knowledge graph of the corresponding field branch according to the extended entity dependency relationship identification result;

Further, displaying auxiliary data which is corresponding to the writing content and needs to be quickly queried in a specified area of a current writing page, so that a user can select and use the auxiliary data in the writing process, and the method comprises the following steps:

rendering auxiliary data which correspond to the written content and need to be quickly queried in different forms to obtain rendering results of texts, tables, charts or rich media of the auxiliary data;

and displaying the rendering result of the text, the table, the chart or the rich media of the auxiliary data in a specified area of the current writing page so as to be selected and used by a user in the writing process according to the rendering result.

In a second aspect, an embodiment of the present invention further provides an intelligent auxiliary writing processing apparatus, including:

the receiving module is used for receiving the writing content which needs to be quickly inquired and is triggered and sent by a user in the writing process of the current writing page in a preset triggering mode;

the first acquisition module is used for performing query intention identification on the writing content of the auxiliary data needing quick query and acquiring a query intention identification result of the writing content of the auxiliary data needing quick query; wherein the query intention identification result comprises a query intention category identification result and an entity dependency relationship identification result; the entity dependency relationship identification result is obtained by depending on the entity identification result and the dependency relationship among the entities;

the second acquisition module is used for inquiring the knowledge graph according to the inquiry intention type identification result and the entity dependence relationship identification result and acquiring auxiliary data which corresponds to the writing content and needs to be quickly inquired;

and the display module is used for displaying the auxiliary data which is corresponding to the writing content and needs to be quickly inquired in a specified area of the current writing page so as to be selected and used by a user in the writing process according to the auxiliary data.

In a third aspect, an embodiment of the present invention further provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, where the processor implements the intelligent assisted writing processing method according to the first aspect when executing the computer program.

In a fourth aspect, an embodiment of the present invention further provides a non-transitory computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the intelligent assisted writing processing method according to the first aspect.

As can be seen from the above technical solutions, the method, the apparatus, the electronic device, and the storage medium for processing the intelligent auxiliary writing provided in the embodiments of the present invention perform query intention recognition on writing content of auxiliary data to be quickly queried, obtain a query intention recognition result of the writing content of the auxiliary data to be quickly queried, further query a knowledge graph according to the query intention type recognition result and the entity dependency relationship recognition result, obtain auxiliary data corresponding to the writing content and requiring quick query, and display the auxiliary data corresponding to the writing content and requiring quick query in a specified area of a current writing page for a user to select and use the auxiliary data in a writing process, so that the embodiments of the present invention can provide intelligent writing assistance in a writing process of the user, and quickly provide the auxiliary data required to be queried in the writing process of the user, the method and the device can save the process that a user searches for auxiliary data by switching the search platform in the writing process, and the key point of the embodiment of the invention is that the query intention of the user about the writing auxiliary data can be analyzed and obtained from the writing content, so that the corresponding writing auxiliary data can be obtained from the related knowledge graph according to the query intention, the writing efficiency and the writing effect can be effectively improved, and the intelligent writing becomes possible. Therefore, according to the intelligent auxiliary writing technical scheme provided by the embodiment of the invention, the auxiliary data which corresponds to the writing content and needs to be quickly inquired can be provided for the user only by triggering the writing content which needs to be quickly inquired of the auxiliary data in the writing process, so that the user can directly utilize the corresponding auxiliary data to assist the writing process, and the writing efficiency and the writing effect can be improved.

Drawings

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

FIG. 1 is a flow chart of a method for intelligent assisted authoring processing provided by an embodiment of the present invention;

FIG. 2 is a block diagram of a authoring platform architecture according to an embodiment of the present invention;

FIG. 3 is a schematic diagram illustrating an implementation process of an intelligent auxiliary writing processing method according to an embodiment of the present invention;

fig. 4, fig. 5, fig. 6, fig. 7, fig. 8 and fig. 9 are schematic diagrams of answer query recommendation results corresponding to the written content, obtained according to the query intention recognition result according to the embodiment of the present invention;

FIG. 10 is a system block diagram of an artificial intelligence based intelligent authoring assistant provided by an embodiment of the present invention;

FIG. 11 is a schematic diagram illustrating steps of a human-computer interaction method based on an intelligent writing assistant according to an embodiment of the present invention;

FIG. 12 is a schematic structural diagram of an intelligent auxiliary writing processing apparatus according to an embodiment of the present invention;

fig. 13 is a schematic structural diagram of an electronic device according to an embodiment of the invention.

Detailed Description

The following further describes embodiments of the present invention with reference to the accompanying drawings. The following examples are only for illustrating the technical solutions of the present invention more clearly, and the protection scope of the present invention is not limited thereby.

Fig. 1 is a flowchart illustrating an intelligent auxiliary writing processing method according to an embodiment of the present invention, and it should be noted that the intelligent auxiliary writing processing method according to the embodiment of the present invention can be implemented based on the writing platform architecture illustrated in fig. 2. The intelligent auxiliary writing processing method provided by the embodiment of the invention is explained and explained in detail with reference to fig. 1. As shown in fig. 1, an intelligent auxiliary writing processing method provided in an embodiment of the present invention specifically includes the following steps:

step 101: and receiving the writing content which is triggered and sent by a user in the writing process of the current writing page in a preset triggering mode and needs to quickly inquire the auxiliary data.

In this step, it is assumed that the user wants to query some auxiliary data during the writing process of the current writing page, for example, the user writes a financial report about couchmate, and at this time, the user writes content about the business income of three quarters before 2019, but since this part belongs to data-related content, the user does not generally remember, so the user needs to query the related content. In the prior art, when the user needs to search the income of business three quarters before 2019 on some search platforms such as a hundredth platform or a dog search platform, and then screens and copies the search results and pastes the search results into the writing process, but the processing method has the defects that the user needs to switch to a hundredth search page and the like to search related contents, and further screening and copying are needed by the user, and the search results are pasted into the writing process, so that the process is very complicated, much time is occupied by the user, the user experience is very bad, and particularly, the user is troublesome to query and screen some complex data. Specifically, as shown in fig. 3, in the writing process of the current writing page, a user may directly trigger writing content (for example, "business income of three quarters before couchmate 2019") written on the current page by the user, which needs to quickly query auxiliary data, to a server through a preset trigger manner (for example, through triggering of a shortcut key Ctr + q), and then the server performs query intention recognition on the writing content (business income of three quarters before couchmate 2019), obtains a query intention recognition result (object: couchmate; time: three quarters before 2019; index: business income), and queries a knowledge graph according to the query intention recognition result, obtains an answer query recommendation result corresponding to the writing content (for example, an accurate answer: business income of three quarters before couchmate 2019), and/or the related answers comprise the business income of the last half year of the couchgrass 2019, the business income of the first quarter of the couchgrass 2019, the business income of the quarter trend of the couchgrass 2019, and the like), and the answer query recommendation result is displayed in a specified area of the current writing page so as to be directly selected and used by the user for the accurate answer and/or the related answer in the writing process. For example, as shown in fig. 7, the user may select the exact answer directly in the designated area: the business income of the couchgrass 2019 in the first three seasons, and/or relevant answers including the business income of the couchgrass 2019 in the last half year, the business income of the couchgrass 2019 in the first quarter, the business income trend of the couchgrass 2019 in the first quarter, and the like, so that the writing process is smoothly completed. In addition, fig. 8 and 9 show that the contents of the composition are: the accurate recommendation result and the related recommendation result of the operating income and net profit of the couchgrass in Guizhou province in three quarters before 2019 are not described in detail.

In addition, in other embodiments of the present invention, the server may further perform different forms of rendering on the accurate answers and/or the related answers, and then display the rendered accurate answers and/or related answers in a specified area of the current writing page. For example, the server may: the operating income of the couchmate three quarters before 2019 is rendered into a chart format and then displayed in the specified area of the current writing page, so that a user can directly use the contents in the chart format from the specified area of the current writing page to solve the problem of matching the chart in the writing process, thereby greatly simplifying the writing process and perfectly assisting in finishing intelligent writing. Without the rendering process, the user needs to draw the relevant table after getting the revenue three quarters of the year before the couchmate 2019, which undoubtedly increases the writing burden and reduces the writing efficiency. It should be noted that the specific rendering manner is specifically described in other embodiments, and is not limited to the chart format, and may also be rendered in the form of text, table, rich media, and the like in a specified format as needed. As also shown with reference to the chart rendering results shown in fig. 4, 5, and 6, the recommendation results are shown as the capital flow direction of the Guizhou couchwood, wherein for fig. 4, 5, and 6, the recommendation results give recommendations in different ways from different dimensions, such as Guizhou couchwood 5-day principal force inflow, Guizhou couchwood 5-day principal force outflow, Guizhou couchwood 5-day principal force net inflow, Guizhou couchwood current-day capital flow direction, Guizhou couchwood current-day capital flow trend chart, and so on.

Step 102: performing query intention recognition on the writing content of the auxiliary data needing to be quickly queried, and acquiring a query intention recognition result of the writing content of the auxiliary data needing to be quickly queried; wherein the query intention identification result comprises a query intention category identification result and an entity dependency relationship identification result; the entity dependency relationship identification result is obtained according to the entity identification result and the dependency relationship among the entities.

In this step, after the user triggers and sends the writing content that needs to be subjected to the auxiliary data query, the writing content needs to be subjected to query intention recognition first to recognize the query intention. Generally, entity dependency identification results are included in the query intent identification results. For example, suppose the written content of the auxiliary data that needs to be quickly queried is: "total business income realized in 2019 of Guizhou thatch is ___ billion yuan" (where "___" represents auxiliary data in written content requiring quick query), query intention recognition can be performed on the written content: the identification results of entities ("company", "time", "index name") are acquired as "Guizhou council station", "2019", "business income", "business profit", respectively, and then the identification results of the dependencies among the entities are acquired as a triple ("Guizhou council station", "2019", "business income"). In addition, when the query intention is identified, the category to which the query intention belongs is generally obtained at the same time, and the field to which the query intention belongs may also be understood as obtaining, for example, determining whether the query intention belongs to financial auxiliary data query, economic auxiliary data query, educational auxiliary data query, or the like, so as to conveniently query the knowledge graph of the corresponding field to obtain the desired auxiliary data.

Step 103: inquiring a knowledge graph according to the inquiry intention type identification result and the entity dependence relationship identification result, and acquiring auxiliary data which corresponds to the writing content and needs to be inquired quickly;

in this step, the knowledge graph of the corresponding domain branch may be determined according to the query intention type identification result, and then the auxiliary data corresponding to the written content and requiring quick query may be acquired from the knowledge graph of the corresponding domain branch according to the entity dependency relationship identification result.

Having an Association relationship step 104: and displaying the auxiliary data which is corresponding to the writing content and needs to be quickly inquired in a specified area of the current writing page so as to be selected and used by a user aiming at the auxiliary data in the writing process.

In this embodiment, after the server queries the auxiliary data that needs to be quickly queried and corresponds to the writing content, the auxiliary data that needs to be quickly queried and corresponds to the writing content is displayed in a specified area of the current writing page (for example, a right area of the current writing page), so that a user can directly drag or paste the auxiliary data located in the specified area in the writing process of the current page for auxiliary writing, thereby greatly facilitating the user. Therefore, the embodiment can provide intelligent writing assistance in the writing process of the user, and omits the processes of searching the auxiliary data and organizing the auxiliary data by switching the search platform in the writing process of the user. Therefore, according to the intelligent auxiliary writing technical scheme provided by the embodiment of the invention, the auxiliary data which corresponds to the writing content and needs to be quickly inquired can be provided for the user only by triggering the writing content which needs to be quickly inquired of the auxiliary data in the writing process, so that the user can directly utilize the corresponding auxiliary data to assist the writing process, and the writing efficiency and the writing effect can be improved.

In this embodiment, it should be noted that a user may connect a text writing page to an intelligent writing robot through a PC terminal or a mobile terminal (IOS system or Android system), the intelligent writing robot is in a standby state during a normal writing process of the user, when the user needs to query a certain item of data when inputting a piece of text, the writing content of the auxiliary data to be queried may be sent to a writing server at a cursor position in a preset trigger manner, the writing server performs intent analysis and query on the writing content of the auxiliary data to be queried, so as to obtain a query result about the auxiliary data, and sends the query result to the text writing page, and the query result is displayed in a designated area in the text writing page for the user to directly use in the writing process. It should be noted that, when the user uses the query result from the designated area, various manners such as dragging, pasting, cutting, copying, etc. may be adopted, which is not limited in this embodiment. For example, query results may also be used from a specified area by way of class input method interactions. For example, the user can insert answer data quickly by keyboard up and down selection, 0-9 quick data, space bar, selection mode.

In this embodiment, it should be noted that, when the writing content of the auxiliary data to be queried is sent to the writing server by a preset trigger mode, the writing content may be triggered by a shortcut key, or may be triggered by a single mouse click or a double mouse click.

In this embodiment, after acquiring the writing content of the auxiliary data that needs to be quickly queried by the user, the writing server performs query intention recognition on the writing content, thereby acquiring a query intention recognition result. It is to be understood that the intention recognition may include recognizing and analyzing writing requirements that may be generated by a user in a current writing scenario, wherein the writing requirements may be multiple. And the query intention recognition result is the writing requirement of the user generated under the current writing scene according to intention recognition.

In this embodiment, the accurate answers and/or the related answers obtained after querying the knowledge graph according to the query intention recognition result are displayed in the designated area of the current writing page, where the designated area may be any blank of the writing page, and is not limited herein. At this time, the user can select and use the auxiliary data for the answer recommendation result written in the blank of the page. Furthermore, the user can insert the accurate answers and/or the related answers of the designated area into various data representation forms such as numerical values, texts, tables, pictures and data modules in an interactive mode such as clicking, selecting, spacing, keyboard shortcut keys and the like.

For example, suppose that the user needs to query the auxiliary data as written content: if the capital flow direction of the Guizhou Hitachi on the same day is changed, firstly, a user triggers the written content 'the capital flow direction of the Guizhou Hitachi on the same day' in a preset triggering mode, and according to the operation triggered by the user, the written content 'the capital flow direction of the Guizhou Hitachi on the same day' is identified by the query intention, so that the query intention identification result is obtained: the day, Guizhou Hitachi, and the flow of funds, it is understood that the query intent recognition results may extract multiple word slot content from the written intent for providing multiple dimensional answer references. Further, acquiring an answer query recommendation result according to a query intention recognition result 'the current day, Guizhou Maotai and capital flow direction' query knowledge graph: capital flow data (accurate answer) of Guizhou Maotai on the same day, capital flow data (related answer) of Guizhou Maotai nearly 5 days, and Guizhou Maotai stock information (related answer) of Guizhou Maotai on the same day.

It can be known from the foregoing technical solutions that, in the intelligent auxiliary authoring processing method provided in the embodiments of the present invention, by performing query intention recognition on authoring content requiring auxiliary data to be quickly queried, a query intention recognition result of the authoring content requiring auxiliary data to be quickly queried is obtained, and then a knowledge graph is queried according to the query intention type recognition result and the entity dependency recognition result, auxiliary data requiring quick querying corresponding to the authoring content is obtained, and the auxiliary data requiring quick querying corresponding to the authoring content is displayed in a specified area of a current authoring page for a user to select and use the auxiliary data in the authoring process, so that the embodiments of the present invention can provide intelligent authoring assistance in the user authoring process, quickly provide auxiliary data that the user needs querying in the authoring process, the method and the device can save the process that a user searches for auxiliary data by switching the search platform in the writing process, and the key point of the embodiment of the invention is that the query intention of the user about the writing auxiliary data can be analyzed and obtained from the writing content, so that the corresponding writing auxiliary data can be obtained from the related knowledge graph according to the query intention, the writing efficiency and the writing effect can be effectively improved, and the intelligent writing becomes possible. Therefore, according to the intelligent auxiliary writing technical scheme provided by the embodiment of the invention, the auxiliary data which corresponds to the writing content and needs to be quickly inquired can be provided for the user only by triggering the writing content which needs to be quickly inquired of the auxiliary data in the writing process, so that the user can directly utilize the corresponding auxiliary data to assist the writing process, and the writing efficiency and the writing effect can be improved.

Based on the content of the foregoing embodiment, in this embodiment, performing query intention recognition on the writing content of the auxiliary data needing quick query, and acquiring a query intention recognition result of the writing content of the auxiliary data needing quick query, includes:

In this embodiment, when performing query intention recognition on the writing content requiring quick query auxiliary data and acquiring a query intention recognition result of the writing content requiring quick query auxiliary data, two parts of contents need to be acquired, one part is to acquire a query intention type recognition result, and the other part is to acquire an entity dependency relationship recognition result. The purpose of obtaining the query intention category identification result is to determine the category to which the query intention belongs, for example, whether the query intention belongs to auxiliary data query in financial category, auxiliary data query in economic category, auxiliary data query in education category, or the like, so as to obtain the desired auxiliary data from the knowledge graph in the corresponding field. The purpose of obtaining the entity dependency relationship identification result is to obtain a real query intention, and since the query intention of the writing content of the auxiliary data to be quickly queried can be clearly reflected in the entity dependency relationship identification result, the entity dependency relationship identification result needs to be obtained, and the entity dependency relationship identification needs to depend on the entity identification result and combination constraints among the entity identification results.

For example, suppose the writing content of the auxiliary data requiring quick query is: "the total income and profit of business realized in 2019 th of Guizhou thatch are ___ hundred million yuan and ___ hundred million yuan respectively", the semantic analysis process is as follows: the intention classification model recognition result is "company financial index", the entity ("company", "time", "index name") recognition results are "Guizhou council", "2019", "income of business", "profit of business", respectively, and the dependency relationship recognition result is the 1 st triple ("Guizhou council", "2019", "income of business"). The 2 nd triplet ("Guizhou Maotai", "2019", "operating profit"), thus, the 2 triplets identified above, together with the intention classification of "company financial index", can be used to find the answer in the knowledge graph of the corresponding field.

Based on the content of the foregoing embodiment, in this embodiment, performing query intention type identification on the writing content that needs to be quickly queried as to obtain a query intention type identification result includes:

In this embodiment, it should be noted that, as described above, when performing query intention type identification on the written content that needs the quick query auxiliary data, a query intention type identification result needs to be obtained. It can be understood that the query intention type identification result is used for characterizing the domain branch corresponding to the writing content needing the quick query auxiliary data, and therefore, the domain knowledge graph corresponding to the auxiliary data needing the quick query in the writing content can be determined through the query intention type identification result.

In order to solve the problem of how to obtain the query intention type identification result, the embodiment identifies the query intention type by adopting an intelligent intention type identification model so as to improve the efficiency and the accuracy of the query intention type identification.

Specifically, the intention type recognition model needs to be trained before being used, and in the training, it can be obtained by training based on a machine learning algorithm using a writing content training sample of each known query intention type as sample input data and using a query intention type corresponding to the writing content of each known query intention type as sample output data.

In this embodiment, it should be noted that the above-mentioned intent category identification model is generated based on machine learning algorithm training according to the writing content of each known query intent category. When the intention type recognition model is generated based on machine learning algorithm training, generally, writing content training samples of known query intention types are used as sample input data, query intention types corresponding to the writing content of the known query intention types are used as sample output data, an initial machine learning model is trained until a model convergence condition is met, and then the intention type recognition model is generated. In this embodiment, a CNN or RNN machine learning model may be employed for model training.

Based on the content of the foregoing embodiment, in this embodiment, the performing entity identification and entity dependency identification on the writing content that needs to quickly query the auxiliary data to obtain an entity identification result and an entity dependency identification result includes:

In this embodiment, the written content is segmented to obtain a plurality of sentences with complete semantics;

for each sentence, the following processing procedure is performed: performing word segmentation and part-of-speech tagging on the sentence to further perform entity identification so as to obtain a key entity (object entity) in the sentence, and performing general entity identification on the sentence so as to at least obtain a time entity, a condition entity, an index entity and the like in the sentence;

and judging the incidence relation among the time entity, the condition entity, the index entity and the object entity through dependency relation extraction, and further determining an entity dependency relation identification result among the entities.

For example, suppose the writing content of the auxiliary data requiring quick query is: "the total income and profit of business realized in 2019 of Guizhou thatch are ___ hundred million yuan and ___ hundred million yuan respectively, and the same ratio increases ___% and ____% respectively", and the semantic analysis process is as follows: the intention classification model recognition result is "company financial index", the entity ("company", "time", "index name") recognition results are "Guizhou council", "2019", "income of business", "profit of business", respectively, and the dependency relationship recognition result is the 1 st triple ("Guizhou council", "2019", "income of business"). The 2 nd triplet ("Guizhou Maotai", "2019", "operating profit"), thus, the 2 triplets identified above, together with the intention classification of "company financial index", can be used to find the answer in the knowledge graph of the corresponding field.

In this embodiment, as shown in fig. 3, when performing semantic analysis on the written content, sentence segmentation may be performed first, and then semantic recognition may be performed on the segmented sentence, including performing information of a knowledge graph of related fields such as part of speech recognition, entity recognition, and relationship recognition, so as to finally recognize a time entity, an object entity, an index entity, and semantic dependency relationships among the entities in the written content.

It is understood that semantic analysis is generally to perform analysis and recognition of different "dimensions" on text contents such as sentences and paragraphs by means of natural language processing, machine learning, and the like, and includes, but is not limited to, analysis and recognition modules such as "word segmentation (word segmentation)", "part of speech tagging", "entity recognition", "time recognition", "condition recognition", "dependency extraction (dependency parsing)", "intention recognition", "semantic dependency analysis", and the like. The specific process is as follows, firstly segmenting a target text (paragraph, long sentence) into a sentence with complete semantics, performing entity recognition on the sentence, such as key entities of 'company organization', 'region', 'character' and the like in the sentence, and then performing further 'generic entity' recognition such as time recognition, condition recognition, index recognition and the like, wherein except 'time' and 'condition', ideally, each entity corresponds to the entity knowledge in a knowledge graph (knowledge graph) one by one, and then judging the association among the 'entities', 'condition', 'time' and the like through 'dependency relationship extraction (dependency syntactic analysis').

It can be understood that semantic extraction and recognition generally includes semantic encoding of characters through a pre-training language model to generate high-latitude semantic vectors, performing sequence labeling and classification tasks such as "entity recognition", "relationship extraction", "intention recognition" and the like by using deep learning models such as Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), long-short memory unit (LSTM), attention-based models (such as transformer and derivative models Bert and XLNet based on the transformer) and the like, and performing extraction and recognition processing by using feature engineering processing of conventional machine learning models such as support vector machine, bayes, hidden markov, conditional random field and the like.

Based on the content of the foregoing embodiment, in this embodiment, querying a knowledge graph according to the query intention type identification result and the entity dependency relationship identification result, and acquiring auxiliary data corresponding to the written content and requiring quick query, includes:

In this embodiment, it should be noted that each domain branch corresponds to a corresponding knowledge graph, and therefore, when actually acquiring auxiliary data corresponding to the written content and requiring quick query, the knowledge graph of the corresponding domain branch needs to be determined according to the query intention type identification result, and then the auxiliary data corresponding to the written content and requiring quick query is acquired from the knowledge graph of the domain branch.

It can be understood that knowledge and time series data of each domain branch are stored in the knowledge graph of each domain branch, the knowledge and time series data include each index information of each object at each time in the corresponding domain branch, and the entity dependency relationship identification result includes a combination constraint relationship among the object entity, the time entity and the index entity, so that auxiliary data corresponding to the written content and needing quick query can be acquired from the knowledge graph of the corresponding domain branch according to the entity dependency relationship identification result, thereby achieving the purpose of intelligent auxiliary writing.

For example, suppose the writing content of the auxiliary data requiring quick query is:

"the total income and profit of business realized in 2019 of Guizhou thatch are ___ hundred million yuan and ___ hundred million yuan respectively, and the same ratio increases ___% and ___% respectively", and the semantic analysis process is as follows: the intention classification model recognition result is "company financial index", the entity attribute ("company", "time", "index name") recognition results are "Guizhou council", "2019", "revenue, and" revenue, respectively, and the dependency identification result is the 1 st triple ("Guizhou council", "2019", "revenue). The 2 nd triple (Guizhou Maotai, 2019 and business profit) is identified, and accordingly, the identification results are classified according to the intention of the company financial index through the 2 identified triples, so that corresponding auxiliary data can be searched in a knowledge graph corresponding to the company financial index, and the purpose of intelligent auxiliary writing is achieved.

In this embodiment, it should be noted that, when acquiring auxiliary data that needs to be quickly queried and corresponds to the written content, accurate auxiliary data that needs to be quickly queried and corresponds to the written content (hereinafter, referred to as an accurate answer) may be acquired, and auxiliary data that is related to the written content and needs to be quickly queried and corresponds to the written content (hereinafter, referred to as a related answer) may also be acquired.

It is to be understood that the exact answer is an answer that matches the query intent recognition result. For example, assume that the query intent recognition result is: if the council is the income of the council three quarters before 2019, the accurate answer is the answer which is exactly matched with the income of the council three quarters before 2019, that is, the accurate answer is: business income data of three quarters before 2019 council.

It is to be understood that the relevant answer refers to an answer having an association relationship with the query intention recognition result. For example, assume that the query intent recognition result is: couchmate (subject), three quarters (time) before 2019, revenue (index), then the relevant answer is one that has an association with the revenue of the three quarters before couchmate 2019. Here, the answer having the incidence relation may be an answer matching any one or two factors of the query object, the query time, and the query index. For example, the relevant answers may be: business income of the first half of couchgrass 2019 (query time is changed, but query object and query time are unchanged), business income of the first quarter of couchgrass 2019 (query time is changed, but query object and query time are unchanged), business income trend of the couchgrass 2019 (query time and query index are changed, but query object is unchanged), business income of the first three quarters of couchgrass 2019 (query object is changed, but query time and query index are unchanged), charitable donation amount of the first three quarters of couchgrass 2019 (query index is changed, but query time and query object are unchanged), stock sales amount of the first three quarters of couchgrass 2019 (query index is changed, but query time and query object are unchanged), and the like.

In this step, it should be noted that the meaning of providing an accurate answer is: the user can be helped to complete writing the required data. And the significance of providing relevant answers is: the method can help the user to disperse or expand the writing thought and the writing content, or help the user to form data contrast in the writing, or provide richer data support for the user in the writing, and increase the attraction and persuasion of the article. In addition, the significance of providing relevant answers is that providing relevant answers can help the user as much as possible when the written content of a certain query does not have an exact matching result in the knowledge graph. Therefore, the embodiment can provide the accurate answers and the related answers for the user at the same time, namely, the embodiment can provide the answers with multiple dimensions for the user to refer to, so that more data selections can be provided for the user, and the user experience is improved.

It can be understood that, in order to provide the relevant answer, in this embodiment, it is necessary to determine, according to the entity dependency identification result, an extended entity dependency identification result corresponding to the entity dependency identification result; wherein, the object entity in the extended entity dependency relationship identification result and the object entity in the entity dependency relationship identification result have a class relationship or a competitive product relationship (such as a relationship between couchgrass platform and wuliangye, or a relationship between seabed fishing and seafishing workshop, or a relationship between Shanghai tobacco and Beijing tobacco, etc.); the time entities in the extended entity dependency identification result and the time entities in the entity dependency identification result have adjacent relations or periodic relations (such as the relation between the first quarter and the second quarter, the relation between the first quarter of the last year and the first quarter of the current year, etc.); the index entity in the extended entity dependency relationship identification result and the index entity in the entity dependency relationship identification result have a similar relationship or an opposite relationship (for example, a relationship between total annual income and total annual profit, a relationship between total annual income and annual loss), it can be understood that, since the knowledge graph of the corresponding domain branch stores knowledge and time series data of the corresponding domain branch, the knowledge and time series data includes each index information of each object in the corresponding domain branch at each time, so as to provide an accurate answer and a related answer at the same time, accurate auxiliary data corresponding to the written content and needing to be quickly queried can be obtained from the knowledge graph of the corresponding domain branch according to the entity dependency relationship identification result; and acquiring related auxiliary data which corresponds to the writing content and needs to be quickly inquired from the knowledge graph of the corresponding field branch according to the extended entity dependency relationship identification result, so that accurate answers and related answers can be provided at the same time for a user to use as required.

Based on the content of the foregoing embodiment, in this embodiment, displaying, in a designated area of a current writing page, auxiliary data that needs to be quickly queried and corresponds to the writing content, so that a user can select and use the auxiliary data in a writing process, where the method includes:

In this embodiment, it can be understood that the obtained auxiliary data that needs to be quickly queried and corresponds to the written content may be displayed in a blank of a writing face in a rendering form of text, table, chart, or rich media, for example, the obtained auxiliary data that needs to be quickly queried and corresponds to the written content is automatically drawn to form a multi-specification graph, which facilitates data insertion and use for a user to select and use the auxiliary data.

In this embodiment, different forms of rendering processing are performed on auxiliary data corresponding to the written content and needing quick query, so that a user can select answers of different rendering results, and the user experience is improved while the answer display form is enriched. The rendering result can be pictures, charts, tables, texts with specified formats and the like, so that various data forms are provided for users, the users can conveniently select the data forms according to needs in the writing process, and the intelligent writing quality and the writing richness can be improved.

In this embodiment, it should be noted that, in the conventional writing method, data and knowledge are mainly queried and retrieved manually, and data such as tables and pictures are repeatedly drawn and organized, which is often tedious and inefficient. With the increasing development of deep learning, high-quality writing scenes (macro economy, macro strategy, industry, analysis reports and weekly reports of companies, company financial and newspaper comments, government reports, academic reports and the like) depend on a large amount of data, and the whole process of title, material collection, writing, drawing, auditing and publishing is realized, so that the production efficiency of articles and reports is low, the quality can be guaranteed, and a lot of people are difficult to get rid of heavy labor. Therefore, the embodiment redefines an intelligent content production mode, and the production of the mode can greatly improve the scale and stability of content production. In this mode, the machine writing is finally simplified into article assembly through a series of prefabricated parts, and the article production mode is industrially modified by using the mode, and the core flow and the quality control of content production are moved to the prefabricated part production link. Compared with the traditional writing tool, the intelligent auxiliary writing processing method provided by the embodiment has the following advantages: the answer recommendation level is as follows: no matter the task answer configuration or the question-answer type answer configuration, the self-learning can be realized by adjusting the dynamic parameters to update in real time, and the functions of self-learning, active reporting and the like are adopted, so that writing support is provided for intention identification and answer repackaging, and the traditional intelligent writing is more dependent on search knowledge and fixed knowledge content. And (3) writing an interactive layer: the method accords with normal writing interaction, and the user experiences the situation that the water reaches the canal, thereby perfectly meeting the interaction requirements of workers. And (3) knowledge management layer: the knowledge-graph technology is an important component of artificial intelligence technology, and describes concepts, entities and the relationship of keys thereof in the objective world in a structured manner. The knowledge graph extraction technology provides a better capacity for organizing, managing and understanding mass information of the internet, and expresses the information of the internet into a form closer to the human cognitive world. Therefore, a knowledge graph with semantic processing capability and open interconnection capability is established, and great application value can be generated in intelligent writing and personalized recommendation.

In this embodiment, as shown in fig. 10, the intelligent writing robot may be connected to a text writing page of a PC or a mobile terminal (IOS or Android), the intelligent writing assistant is in a standby state during a normal writing process of a user, when the user needs to query a certain item of data when inputting a text, the user query text may be input to the writing server by triggering a shortcut key such as Ctrl + Q at a cursor position, the writing server performs sentence segmentation processing on the input text, extracts key semantic information in the text, identifies a user true intention of the text block, performs reference disambiguation and complex entity identification, and finally constructs the text into a semantic analysis result. And searching corresponding data and data rendering information by using the semantic analysis result, returning the data to the multi-type writing template output module, and finally returning the data to the user according to a rendering format for clicking and inserting the data into the current writing area. As shown in fig. 11, the steps of the human-computer interaction method based on the intelligent writing robot are as follows: step 1: the intelligent writing assistant text entry page is opened. Step 2: receiving text content before a cursor when a user presses a shortcut key (such as Ctrl + Q). And step 3: and sending the received text content to the intelligent writing server. And 4, step 4: the intelligent writing service analyzes the text information and outputs a semantic analysis result. And 5: and querying corresponding data and a rendering mode by using the semantic analysis result. Step 6: and presenting the inquired data to the user in a rendering mode. And 7: the user selects and clicks on the desired data content, inserting it at the cursor location.

Based on the same inventive concept, another embodiment of the present invention provides an intelligent auxiliary writing processing apparatus, and referring to fig. 12, an embodiment of the present invention provides a structural schematic diagram of the intelligent auxiliary writing processing apparatus, where the intelligent auxiliary writing processing apparatus includes: a receiving module 21, a first obtaining module 22, a second obtaining module 23 and a display module 24, wherein:

Since the intelligent auxiliary writing processing apparatus provided in this embodiment can be used to execute the intelligent auxiliary writing processing method described in the above embodiment, and the operation principle and the beneficial effect are similar, detailed descriptions will not be provided here, and specific contents can be referred to the introduction and description of the above embodiment.

Based on the same inventive concept, another embodiment of the present invention provides an electronic device, which is described with reference to fig. 13 and specifically includes the following contents: a processor 301, a memory 302, a communication interface 303, and a communication bus 304;

the processor 301, the memory 302 and the communication interface 303 complete mutual communication through the communication bus 304; the communication interface 303 is used for realizing information transmission between the devices;

the processor 301 is configured to call a computer program in the memory 302, and the processor implements all the steps of the above-mentioned intelligent assisted writing processing method when executing the computer program, for example, the processor implements the following steps when executing the computer program: receiving writing content which is triggered and sent by a user in the writing process of the current writing page in a preset triggering mode and needs to quickly inquire auxiliary data; performing query intention recognition on the writing content of the auxiliary data needing to be quickly queried, and acquiring a query intention recognition result of the writing content of the auxiliary data needing to be quickly queried; wherein the query intention identification result comprises a query intention category identification result and an entity dependency relationship identification result; the entity dependency relationship identification result is obtained by depending on the entity identification result and the dependency relationship among the entities; inquiring a knowledge graph according to the inquiry intention type identification result and the entity dependence relationship identification result, and acquiring auxiliary data which corresponds to the writing content and needs to be inquired quickly; and displaying the auxiliary data which is corresponding to the writing content and needs to be quickly inquired in a specified area of the current writing page so as to be selected and used by a user aiming at the auxiliary data in the writing process.

Based on the same inventive concept, yet another embodiment of the present invention provides a non-transitory computer-readable storage medium, having stored thereon a computer program, which when executed by a processor implements all the steps of the above-mentioned user comment generating method, for example, the processor implements the following steps when executing the computer program: receiving writing content which is triggered and sent by a user in the writing process of the current writing page in a preset triggering mode and needs to quickly inquire auxiliary data; performing query intention recognition on the writing content of the auxiliary data needing to be quickly queried, and acquiring a query intention recognition result of the writing content of the auxiliary data needing to be quickly queried; wherein the query intention identification result comprises a query intention category identification result and an entity dependency relationship identification result; the entity dependency relationship identification result is obtained by depending on the entity identification result and the dependency relationship among the entities; inquiring a knowledge graph according to the inquiry intention type identification result and the entity dependence relationship identification result, and acquiring auxiliary data which corresponds to the writing content and needs to be inquired quickly; and displaying the auxiliary data which is corresponding to the writing content and needs to be quickly inquired in a specified area of the current writing page so as to be selected and used by a user aiming at the auxiliary data in the writing process.

In addition, the logic instructions in the memory may be implemented in the form of software functional units and may be stored in a computer readable storage medium when sold or used as a stand-alone product. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.

The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the embodiment of the present invention. One of ordinary skill in the art can understand and implement it without inventive effort.

Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. Based on such understanding, the above technical solutions may be embodied in the form of a software product, which may be stored in a computer-readable storage medium, such as a ROM/RAM, a magnetic disk, an optical disk, etc., and includes several instructions for enabling a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the user comment generating method according to the various embodiments or some parts of the embodiments.

Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.

Claims

1. An intelligent auxiliary writing processing method is characterized by comprising the following steps:

2. The intelligent auxiliary writing processing method according to claim 1, wherein performing query intention recognition on the writing content requiring quick query auxiliary data to obtain a query intention recognition result of the writing content requiring quick query auxiliary data includes:

3. The intelligent auxiliary writing processing method according to claim 2, wherein the step of performing query intention type recognition on the writing content requiring the quick query auxiliary data to obtain a query intention type recognition result comprises:

4. The intelligent auxiliary writing processing method according to claim 2, wherein performing entity recognition and entity dependency recognition on the writing content requiring the quick query auxiliary data to obtain an entity recognition result and an entity dependency recognition result includes:

5. The intelligent auxiliary writing processing method according to claim 3, wherein querying a knowledge graph according to the query intention type recognition result and the entity dependency relationship recognition result to obtain auxiliary data corresponding to the writing content and requiring quick query, includes:

6. The intelligent auxiliary writing processing method according to claim 3, wherein querying a knowledge graph according to the query intention type recognition result and the entity dependency relationship recognition result to obtain auxiliary data corresponding to the writing content and requiring quick query, includes:

7. The intelligent auxiliary writing processing method according to claim 5 or 6, wherein the step of displaying auxiliary data corresponding to the writing content and requiring quick query in a designated area of a current writing page for a user to select and use the auxiliary data in the writing process comprises:

8. An intelligent assisted authoring processing apparatus, comprising:

9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the steps of the intelligent assisted authoring process of any of claims 1 to 7 when executing the program.

10. A non-transitory computer-readable storage medium, on which a computer program is stored, the computer program, when being executed by a processor, implementing the steps of the intelligent assisted authoring processing method of any one of claims 1 to 7.