CN116383508A - Searching method, searching device, computer equipment and storage medium - Google Patents

Searching method, searching device, computer equipment and storage medium Download PDF

Info

Publication number
CN116383508A
CN116383508A CN202310383445.1A CN202310383445A CN116383508A CN 116383508 A CN116383508 A CN 116383508A CN 202310383445 A CN202310383445 A CN 202310383445A CN 116383508 A CN116383508 A CN 116383508A
Authority
CN
China
Prior art keywords
function
network model
training
search
training sample
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202310383445.1A
Other languages
Chinese (zh)
Inventor
张萍
李积宏
边露
王彩霞
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Industrial and Commercial Bank of China Ltd ICBC
Original Assignee
Industrial and Commercial Bank of China Ltd ICBC
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Industrial and Commercial Bank of China Ltd ICBC filed Critical Industrial and Commercial Bank of China Ltd ICBC
Priority to CN202310383445.1A priority Critical patent/CN116383508A/en
Publication of CN116383508A publication Critical patent/CN116383508A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/35—Clustering; Classification
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9538—Presentation of query results
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/30—Semantic analysis
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

The present application relates to the field of artificial intelligence technology, and relates to a search method, apparatus, computer device, storage medium and computer program product, where the method is applied to applications integrating multiple platform functions. The method comprises the following steps: receiving search sentences input by a user, inputting the search sentences into a functional recognition network model after training, processing the search sentences according to the functional recognition network model, and determining the confidence level of each functional label, wherein the functional labels comprise functional information and platform information. And determining a display sequence according to the confidence degree of each function label, and displaying recommended function entries corresponding to each function label based on the display sequence. By adopting the method, the searching accuracy can be improved.

Description

Searching method, searching device, computer equipment and storage medium
Technical Field
The present application relates to the field of artificial intelligence technology, and in particular, to a search method, apparatus, computer device, storage medium, and computer program product.
Background
With the development of application technology, some applications may integrate multiple platform functions, such as functions of the application itself, functions of a third party application, functions of a micro application, and so on. Taking a bank as an example, the bank itself has some functions for providing users, such as transfer, balance inquiry, etc., and also has functions for accessing third party applications, such as electric charge, gas charge, etc., and also has functions for accessing some micro applications.
In the related art, when a user performs function search in an application integrating functions of a plurality of platforms, the application generally performs function entry recommendation by using keywords input by the user, and the functions of the plurality of platforms are crossed to enable the search words to be crossed, so that the corresponding function entries cannot be accurately given out based on the keywords input by the user, the user wants to search the function of the platform A, but the search result is not very accurate and cannot meet the search intention of the user according to the recommendation entering the function of the platform B.
Disclosure of Invention
In view of the foregoing, it is desirable to provide a search method, apparatus, computer device, computer-readable storage medium, and computer program product that can improve the accuracy of search results.
In a first aspect, the present application provides a search method, where the method is applied to an application integrating multiple platform functions, the method including:
receiving a search statement input by a user;
inputting the search statement into a functional identification network model with the training completed;
processing the search statement according to the function identification network model, and determining the confidence coefficient of each function label; the function tag comprises function information and platform information;
And determining a display sequence according to the confidence coefficient of each function label, and displaying a recommended function entry corresponding to each function label based on the display sequence.
In one embodiment, the determining a display order according to the confidence level of each function label, and displaying the recommended function entry corresponding to each function label based on the display order includes:
under the condition that the current historical behavior data of the user reaches a preset condition, determining the priority of each function label based on the current historical behavior data of the user;
determining the display sequence of each function label based on the priority of each function label and the confidence of each function label, and displaying the recommended function entry corresponding to each function label based on the display sequence of each function label.
In one embodiment, the method further comprises:
updating the training sample set under the condition that an updating instruction is detected; the training sample set consists of a plurality of search statement training samples marked with function labels;
acquiring a current training sample set;
inputting the current training sample set to an initial function identification network model, and training the initial function identification network model;
Under the condition that training is carried out until the initial function identification network model meets a preset loss condition, determining that the initial function identification network model training is completed;
and taking the initial functional identification network model after training as the functional identification network model after training.
In one embodiment, the updating the training sample set when the update instruction is detected includes:
under the condition that the update instruction is detected to be a new first function label instruction, acquiring a search statement training sample marked with the first function label, adding the search statement training sample of the first function label into the training sample set, and updating the training sample set.
In one embodiment, the updating the training sample set when the update instruction is detected includes:
and deleting the search statement training sample marked with the second function label in the training sample set under the condition that the updating instruction is detected to be the second function label deleting instruction, and updating the training sample set.
In one embodiment, the obtaining the current training sample set includes:
under the condition that the preset moment is reached, acquiring a current training sample set; the preset time is a period arrival time or a preset time.
In one embodiment, after taking the trained initial function recognition network model as the trained function recognition network model, the method further comprises:
releasing the trained function recognition network model to a search interface;
the step of inputting the search statement into the trained function recognition network model comprises the following steps:
and calling the functional identification network model which is completed through training through a search interface, and inputting the search statement into the functional identification network model.
In one embodiment, the inputting the search statement into the trained functional recognition network model includes:
the query function identifies whether the training of the network model is completed;
and under the condition that the training of the functional identification network model is completed, inputting the search statement into the functional identification network model with the completed training.
In a second aspect, the present application further provides a search apparatus, where the apparatus is applied to an application integrating multiple platform functions, the apparatus including:
the receiving module is used for receiving search sentences input by a user;
the input module is used for inputting the search statement into the functional identification network model after training;
The determining module is used for processing the search statement according to the function identification network model and determining the confidence coefficient of each function label; the function tag comprises function information and platform information;
and the display module is used for determining a display sequence according to the confidence coefficient of each function label and displaying recommended function entries corresponding to each function label based on the display sequence.
In one embodiment, the display module specifically includes:
a determining unit, configured to determine, based on current historical behavior data of the user, a priority of each of the function tags if it is determined that the current historical behavior data of the user reaches a preset condition;
and the display unit is used for determining the display sequence of each function label based on the priority of each function label and the confidence of each function label and displaying the recommended function entry corresponding to each function label based on the display sequence of each function label.
In one embodiment, the apparatus further includes:
the updating module is used for updating the training sample set under the condition that the updating instruction is detected; the training sample set consists of a plurality of search statement training samples marked with function labels;
The acquisition module is used for acquiring a current training sample set;
the training module is used for inputting the current training sample set into an initial function identification network model and training the initial function identification network model;
the training completion module is used for determining that the initial function identification network model training is completed under the condition that the initial function identification network model meets the preset loss condition after training;
and the function recognition network model determining module is used for taking the initial function recognition network model after training as the function recognition network model after training.
In one embodiment, the update module is specifically configured to:
under the condition that the update instruction is detected to be a new first function label instruction, acquiring a search statement training sample marked with the first function label, adding the search statement training sample of the first function label into the training sample set, and updating the training sample set.
In one embodiment, the update module is specifically configured to:
and deleting the search statement training sample marked with the second function label in the training sample set under the condition that the updating instruction is detected to be the second function label deleting instruction, and updating the training sample set.
In one embodiment, the acquiring module is specifically configured to:
under the condition that the preset moment is reached, acquiring a current training sample set; the preset time is a period arrival time or a preset time.
In one embodiment, the apparatus further comprises:
the issuing module is used for issuing the trained function identification network model to the search interface;
at this time, the input module is specifically configured to:
and calling the functional identification network model which is completed through training through a search interface, and inputting the search statement into the functional identification network model.
In one embodiment, the input module specifically includes:
the query unit is used for querying whether the training of the function recognition network model is finished;
and the input unit is used for inputting the search statement into the functional identification network model after the training is completed under the condition that the training of the functional identification network model is completed.
In a third aspect, the present application also provides a computer device. The computer device comprises a memory storing a computer program and a processor which when executing the computer program performs the steps of:
Receiving a search statement input by a user;
inputting the search statement into a functional identification network model with the training completed;
processing the search statement according to the function identification network model, and determining the confidence coefficient of each function label; the function tag comprises function information and platform information;
and determining a display sequence according to the confidence coefficient of each function label, and displaying a recommended function entry corresponding to each function label based on the display sequence.
In a fourth aspect, the present application also provides a computer-readable storage medium. The computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of:
receiving a search statement input by a user;
inputting the search statement into a functional identification network model with the training completed;
processing the search statement according to the function identification network model, and determining the confidence coefficient of each function label; the function tag comprises function information and platform information;
and determining a display sequence according to the confidence coefficient of each function label, and displaying a recommended function entry corresponding to each function label based on the display sequence.
In a fifth aspect, the present application also provides a computer program product. The computer program product comprises a computer program which, when executed by a processor, implements the steps of:
receiving a search statement input by a user;
inputting the search statement into a functional identification network model with the training completed;
processing the search statement according to the function identification network model, and determining the confidence coefficient of each function label; the function tag comprises function information and platform information;
and determining a display sequence according to the confidence coefficient of each function label, and displaying a recommended function entry corresponding to each function label based on the display sequence.
According to the searching method, the searching device, the computer equipment, the storage medium and the computer program product, semantic analysis is carried out on search sentences input by a user through the model, confidence levels of the user intention to search each platform comprising each function label are further given, recommended function inlets corresponding to each function label are displayed based on the confidence levels of the function labels, on one hand, the accuracy of search results is improved, and on the other hand, the searching efficiency of the user is improved.
Drawings
FIG. 1 is a flow diagram of a search method in one embodiment;
FIG. 2 is a flow chart of a search method according to another embodiment;
FIG. 3 is a block diagram of a search apparatus in one embodiment;
fig. 4 is an internal structural diagram of a computer device in one embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application will be further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the present application.
With the development of application technology, some applications may integrate multiple platform functions, such as functions of the application itself, functions of a third party application, functions of a micro application, and so on.
In the related art, when a user performs function search in an application integrating functions of a plurality of platforms, the application generally performs function entry recommendation by using keywords input by the user, and the functions of the plurality of platforms are crossed to enable the search words to be crossed, so that the corresponding function entries cannot be accurately given out based on the keywords input by the user, the user wants to search the function of the platform A, but the search result is not very accurate and cannot meet the search intention of the user according to the recommendation entering the function of the platform B.
Based on the above, the application provides a searching method, which is applied to applications integrating multiple platform functions. The method comprises the steps of receiving search sentences input by a user, inputting the search sentences into a functional identification network model with complete training, processing the search sentences according to the functional identification network model, and determining the confidence level of each functional label, wherein the functional labels comprise functional information and platform information. And determining a display sequence according to the confidence degree of each function label, and displaying recommended function entries corresponding to each function label based on the display sequence.
According to the searching method, semantic analysis is carried out on the search sentences input by the user through the model, the confidence coefficient of each platform including each function label is given out, the recommended function inlets corresponding to each function label are displayed based on the confidence coefficient of each function label, on one hand, the accuracy of the search results is improved, and on the other hand, the searching efficiency of the user is improved.
In one embodiment, as shown in fig. 1, a search method is provided, where the method is applied to a terminal to illustrate, it is understood that the method may also be applied to a server, and may also be applied to a system including the terminal and the server, and implemented through interaction between the terminal and the server. In this embodiment, the method includes the steps of:
Step 101, receiving a search statement input by a user.
Specifically, the user generally inputs a search term in the search box to search for a corresponding function entry, so the terminal may obtain the search term input by the user after detecting the search instruction input by the user, for example, detecting that the user clicks the search control, or detecting that the user inputs the enter key, etc., to determine the search instruction input by the user, and obtain the search term input by the user to the search box.
In one embodiment, if the method is applied to a server, after the terminal acquires the search statement input by the user, the acquired search statement is sent to the server, and the server receives the search statement sent by the terminal to obtain the search statement input by the user.
Step 103, inputting the search statement into the functional identification network model after training.
The function recognition network model takes a search statement as input and the confidence degree of each function label as output.
Specifically, after the terminal acquires the search statement, the search statement is input into a functional recognition network model after training is completed, and the functional recognition network model classifies the search statement after word segmentation, part-of-speech tagging, syntactic analysis and semantic relation analysis.
And 105, processing the search statement according to the function identification network model, and determining the confidence coefficient of each function label.
The function tag comprises function information and platform information. The confidence is used for representing the matching degree of the function recognition network model prediction search statement and each function label, and the higher the confidence is, the higher the matching degree of the function label and the search statement is.
Specifically, the terminal search sentence is input to the function recognition network model, after word segmentation, part-of-speech tagging, syntactic analysis and semantic relation analysis are performed on the search sentence input by the function recognition network model, the matching degree of the search sentence and each function label is determined, and the confidence degree that the function recognition network model judges the search sentence as each function label is obtained.
And 107, determining a display sequence according to the confidence degree of each function label, and displaying recommended function entries corresponding to each function label based on the display sequence.
Specifically, after determining the confidence level of each function label, the terminal may determine the display order of each function label according to the confidence level of each function label, where the higher the confidence level, the higher the display order. The terminal can also determine the display sequence of each function label according to the confidence level of each function label and the priority of each function label determined by combining the historical behavior information of the user. After determining the display sequence of each function label, the terminal displays the recommended function entry corresponding to each function label based on the display sequence, for example, displays the recommended function entry corresponding to each function label in a list mode, or displays the recommended function entry corresponding to each function label in a left-to-right mode.
In addition, since there may be many function tags included in the application, some recommended function entries corresponding to the function tags are not necessarily displayed, and it is long to display recommended function entries corresponding to all the function tags, in one embodiment, when displaying recommended function entries corresponding to the function tags based on the display order, the terminal may display only recommended function entries corresponding to function tags preset before the display order, for example, display recommended function entries corresponding to the first 5 function tags.
In the embodiment, semantic analysis is performed on search sentences input by a user through the model, confidence levels of each functional label contained in each platform which the user intends to search are further given, recommended function entries corresponding to each functional label are displayed based on the confidence levels of each functional label, on one hand, the accuracy of search results is improved, and on the other hand, the user search efficiency is improved.
In one embodiment, after the terminal determines the confidence coefficient of each function label, the function label with the confidence coefficient higher than a preset confidence coefficient threshold value is used as the function label to be displayed, the confidence coefficient threshold value is used for dividing the degree that each function label predicted by the function recognition network model accords with the mind of a user, and the function label with the confidence coefficient not higher than the preset confidence coefficient threshold value indicates that the function label obviously does not accord with a search statement and is not displayed; the function labels with the confidence degrees higher than the preset confidence degree threshold value indicate that the function labels possibly accord with search sentences and need to be displayed for selection by a user, then the display sequence is determined according to the confidence degrees of the function labels to be displayed, and recommended function entries corresponding to the function labels are displayed based on the display sequence.
In one embodiment, the step 107 specifically includes:
step 107A, determining the priority of each function label based on the current historical behavior data of the user when the current historical behavior data of the user is determined to reach the preset condition.
In particular, some users may not use applications frequently and thus do not accumulate enough historical behavior data; some users may often use the application program, so that enough historical behavior data may be accumulated, and therefore, after the terminal receives a search statement of the user, the terminal may query whether the current historical behavior data of the user reaches a preset condition, for example, whether the search frequency reaches a preset search frequency threshold, or whether the page browsing duration reaches a preset duration threshold, or the like. After the terminal inquires that the current historical behavior data of the user reaches the preset condition, the priority of each function label is determined based on the current historical behavior data of the user, for example, the priority of each function label is determined based on the ratio of the corresponding behavior data of each function label to the total behavior data.
Step 107B, determining a display order of each function label based on the priority of each function label and the confidence of each function label, and displaying the recommended function entry corresponding to each function label based on the display order of each function label.
Specifically, after obtaining the priority of each function label and the confidence coefficient of each function label, the terminal may determine the display sequence of each function label according to the priority of each function label and the confidence coefficient of each function label according to a preset weight relationship.
For example, according to the priority order of each function label, the terminal respectively gives different action weight values to each function label by using a preset action weight distribution strategy, for example, the function label with the highest limit is given a action weight value of 1, the priority is reduced by 0.05 per first stage, and the like until the action weight value is reduced to 0; according to the confidence level of each function label, different search weight values are respectively given to each function label by utilizing a preset search weight distribution strategy, and the example is not continued here like a behavior weight distribution strategy. After obtaining the behavior weight value and the search weight value of each function label, the terminal determines the display weight of the function label according to the behavior weight value and the search weight value of each function label, and determines the display sequence of each function label according to the sequence from large to small of the display weight.
In this embodiment, after the confidence level of each function tag is determined according to the search statement input by the user, the priority level of each function tag is determined in combination with the behavior data of the user, and the display sequence of each function tag is determined according to the priority level and the confidence level of each function tag, so that the display sequence of the recommended function entry comprehensively considers the search intention and the habit of the user, and accords with the mind of the user.
In one embodiment, the method further comprises:
step 109, updating the training sample set when the update instruction is detected.
The training sample set consists of a plurality of search statement training samples marked with function labels.
Specifically, with the deletion of functions of each platform of the application integration, the training sample set needs to be updated accordingly, so that the prediction accuracy of the trained function recognition network model is higher. Therefore, the terminal updates the training sample set according to the update instruction under the condition that the terminal detects the update instruction.
Step 111, acquiring a current training sample set.
Specifically, the terminal may read the training sample set obtained by the preprocessing as a training sample set for training the function recognition network model.
It should be noted that, updating the training sample set and acquiring the current training sample set are not performed synchronously, and the terminal updates the training sample set under the condition that an update instruction is detected; the terminal may acquire the training sample set to train the function recognition network model periodically, or may acquire the training sample set to train the function recognition network model when receiving an instruction to update the function recognition network model.
And 113, inputting the current training sample set into the initial function identification network model, and training the initial function identification network model.
Specifically, after the terminal obtains the training sample set, for each training sample in the training sample set, a search sentence of the training sample is input into an initial function recognition network model, and model parameters in the initial function recognition network model, such as iteration step length, semantic vector size, learning rate and the like, are adjusted according to the output of the initial function recognition network model and the function labels in the disciplinary samples.
And 115, under the condition that training is carried out until the initial function identification network model meets the preset loss condition, determining that the training of the initial function identification network model is completed.
Specifically, the terminal performs iterative training on the initial function recognition network model based on the training sample set until the initial function recognition network model meets a preset loss condition, for example, a part of training samples in the training sample set are utilized to verify the loss of the initial function recognition network model, and the initial function recognition network model is determined to meet the preset loss condition. And under the condition that the initial function recognition network model meets the preset loss condition after training, the terminal determines that the initial function recognition network model training is completed.
Step 117, taking the initial functional identification network model after training as the functional identification network model after training.
Specifically, after the terminal determines that the training of the initial function recognition network model is completed, the initial function recognition network model after the training is completed is used as the function recognition network model after the training is completed, namely, the initial function recognition network model is used as the function recognition network model which is put into use.
In this embodiment, the functional identification network model is trained using a training sample set.
In one embodiment, the step 109 specifically includes:
under the condition that the update instruction is detected to be an newly added first function label instruction, acquiring a search sentence training sample marked with the first function label, adding the search sentence training sample of the first function label into a training sample set, and updating the training sample set.
Specifically, after detecting an update instruction input by an operation and maintenance person of the model, for example, detecting that the operation and maintenance person clicks a control for updating a training sample set, or detecting an instruction for updating the training sample set input by the operation and maintenance person, determining the type of the update instruction, if the update instruction is detected to be a new first function label instruction, collecting search sentences when a user selects a function entry corresponding to the first function label, generating search sentence training samples marked with the first function label, and then adding the search sentence training samples of the first function label into the training sample set to finish updating the training sample set.
In this embodiment, when the application accesses a new function of another platform, a training sample of the new function is collected, and the collected training sample is added to the training sample set to update the training sample set, so that the function recognition network model obtained by training based on the updated training sample set can recognize the new function based on the search statement.
In one embodiment, the step 109 specifically includes:
and deleting the search statement training sample marked with the second function label in the training sample set under the condition that the updating instruction is detected to be the instruction for deleting the second function label, and updating the training sample set.
Specifically, after detecting an update instruction input by an operation and maintenance person of the model, for example, detecting that the operation and maintenance person clicks a control for updating the training sample set, or detecting an instruction for updating the training sample set input by the operation and maintenance person, determining the type of the update instruction, traversing the current training sample set if the update instruction is detected to be an instruction for deleting the second function label, deleting search statement training samples marked with the second function label in the training sample set, and completing updating of the training sample set.
In this embodiment, when an application deletes a function, the training samples corresponding to the function are deleted from the training sample set, and the influence of the deleted training samples on the training of the function identification network model is reduced.
In one embodiment, the step 111 specifically includes:
and under the condition that the preset moment is reached, acquiring a current training sample set.
The preset time is a period arrival time or a preset time.
Specifically, in order to ensure accuracy of the function recognition network model, the function recognition network model may be trained periodically by using a current training sample set, for example, the function recognition network model is trained once a week, and if the time of the first training is 2023, 3, 1, 00, then the next preset time is 2023, 3, 8, 00, and the current training sample set is obtained and trained. The preset time set by the operator can also be, for example, 12 early morning points on sunday, the frequency of using the function identification network model is low, and at the moment, the function identification network model can be retrained through the current training sample set to update the function identification network model, so that 12 early morning points on sunday are set as the preset time.
In this embodiment, the functional identification network model is trained periodically to make the functional identification network model more compatible with the current application scenario.
In one embodiment, the method further comprises:
And issuing the trained function recognition network model to a search interface.
Specifically, the terminal trains the initial functional network model to obtain a trained functional identification network model, and issues the trained functional identification network model to the search interface so that the search process or the user terminal invokes the search interface to use the functional identification network model.
At this time, step 103 specifically includes:
and calling the functional identification network model which is completed through the training through a search interface, and inputting the search statement into the functional identification network model.
Specifically, after receiving the search statement, the terminal invokes the trained function recognition network model through the search interface, and inputs the search statement into the function recognition network model to obtain the confidence coefficient of each function label.
In this embodiment, the model release mode is used for multiple terminals or multiple processes to use one model, so that the loss of computing resources is reduced.
In one embodiment, the step 103 specifically includes:
step 103A, the query function identifies whether the network model is trained.
Specifically, before using the function recognition network model, the terminal inquires whether the function recognition network model is trained, if not, the terminal indicates that the current function recognition network model is unavailable, and a searching mode in the related technology is used for displaying a recommended function entry.
Step 103B, when the training of the functional identification network model is completed, inputting a search sentence into the functional identification network model after the training is completed.
Specifically, if the function recognition network model is trained, indicating that the current function recognition network model is available, inputting a search sentence into the trained function recognition network model to obtain the confidence of each function label.
In this embodiment, before the matching degree between the search statement and each function label is predicted by using the function recognition network model, whether the function recognition network model is trained is queried, so that the prediction accuracy is low due to the fact that an untrained function network recognition model is not used is avoided.
Next, a detailed description of an embodiment provided in the present application is shown in fig. 2, which is a schematic flow chart of a search method according to an embodiment of the present application.
The corpus processing platform is used for receiving and processing training samples, acquiring the training samples of each function of each platform of the user, and generating the training samples consisting of function labels and search sentences based on the historical search sentences and the finally selected functions of each user. In addition, the training samples can be updated according to the functions of adding and deleting the application programs:
In step 201, when an update instruction is detected, an update mode is determined.
Step 202, deleting a training sample corresponding to the first function label in the training sample set when the update instruction is a deletion instruction.
Step 203, adding the training sample corresponding to the second function label to the training sample set when the update instruction is the increase instruction.
The corpus processing platform stores the obtained training sample set so as to enable the training platform to obtain the training sample set.
The training processing platform periodically acquires a training sample set and trains an initial network function model:
step 204, obtaining a current training sample set.
Step 205, inputting the current training sample set to the initial function recognition network model, training the initial function recognition network model, determining that the initial function recognition network model is trained when the initial function recognition network model meets the preset loss condition, taking the trained initial function recognition network model as the trained function recognition network model, and issuing the trained function recognition network model.
The terminal displays the function labels according to the search of the user:
step 206, receiving a search statement input by a user.
Step 207, invoking the function recognition network model issued by the training processing platform, and inputting the search statement into the trained function recognition network model.
And step 208, processing the search statement according to the function identification network model, and determining the confidence level of each function label.
Step 209, determining a display sequence according to the confidence level of each function label, and displaying the recommended function entry corresponding to each function label based on the display sequence.
It should be understood that, although the steps in the flowcharts related to the embodiments described above are sequentially shown as indicated by arrows, these steps are not necessarily sequentially performed in the order indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least some of the steps in the flowcharts described in the above embodiments may include a plurality of steps or a plurality of stages, which are not necessarily performed at the same time, but may be performed at different times, and the order of the steps or stages is not necessarily performed sequentially, but may be performed alternately or alternately with at least some of the other steps or stages.
Based on the same inventive concept, the embodiment of the application also provides a searching device for realizing the searching method. The implementation of the solution provided by the device is similar to the implementation described in the above method, so the specific limitation in one or more embodiments of the search device provided below may refer to the limitation of the search method hereinabove, and will not be repeated herein.
In one embodiment, as shown in fig. 3, there is provided a search apparatus applied to an application integrating a multi-platform function, the apparatus comprising:
a receiving module 301, configured to receive a search sentence input by a user;
an input module 303, configured to input the search statement into a trained functional recognition network model;
a determining module 305, configured to process the search statement according to the function identification network model, and determine a confidence level of each function label; the function tag comprises function information and platform information;
the display module 307 is configured to determine a display order according to the confidence level of each function label, and display a recommended function entry corresponding to each function label based on the display order.
In one embodiment, the display module 307 specifically includes:
a determining unit, configured to determine, based on current historical behavior data of the user, a priority of each of the function tags if it is determined that the current historical behavior data of the user reaches a preset condition;
and the display unit is used for determining the display sequence of each function label based on the priority of each function label and the confidence of each function label and displaying the recommended function entry corresponding to each function label based on the display sequence of each function label.
In one embodiment, the apparatus further includes:
the updating module is used for updating the training sample set under the condition that the updating instruction is detected; the training sample set consists of a plurality of search statement training samples marked with function labels;
the acquisition module is used for acquiring a current training sample set;
the training module is used for inputting the current training sample set into an initial function identification network model and training the initial function identification network model;
the training completion module is used for determining that the initial function identification network model training is completed under the condition that the initial function identification network model meets the preset loss condition after training;
And the function recognition network model determining module is used for taking the initial function recognition network model after training as the function recognition network model after training.
In one embodiment, the update module is specifically configured to:
under the condition that the update instruction is detected to be a new first function label instruction, acquiring a search statement training sample marked with the first function label, adding the search statement training sample of the first function label into the training sample set, and updating the training sample set.
In one embodiment, the update module is specifically configured to:
and deleting the search statement training sample marked with the second function label in the training sample set under the condition that the updating instruction is detected to be the second function label deleting instruction, and updating the training sample set.
In one embodiment, the acquiring module is specifically configured to:
under the condition that the preset moment is reached, acquiring a current training sample set; the preset time is a period arrival time or a preset time.
In one embodiment, the apparatus further comprises:
the issuing module is used for issuing the trained function identification network model to the search interface;
At this time, the input module 303 is specifically configured to:
and calling the functional identification network model which is completed through training through a search interface, and inputting the search statement into the functional identification network model.
In one embodiment, the input module 303 specifically includes:
the query unit is used for querying whether the training of the function recognition network model is finished;
and the input unit is used for inputting the search statement into the functional identification network model after the training is completed under the condition that the training of the functional identification network model is completed.
The various modules in the search apparatus described above may be implemented in whole or in part by software, hardware, and combinations thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.
In one embodiment, a computer device is provided, which may be a terminal, and the internal structure of which may be as shown in fig. 4. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The communication interface of the computer device is used for carrying out wired or wireless communication with an external terminal, and the wireless mode can be realized through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. The computer program is executed by a processor to implement a search method. The display screen of the computer equipment can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, can also be keys, a track ball or a touch pad arranged on the shell of the computer equipment, and can also be an external keyboard, a touch pad or a mouse and the like.
Those skilled in the art will appreciate that the structures shown in FIG. 4 are block diagrams only and do not constitute a limitation of the computer device on which the present aspects apply, and that a particular computer device may include more or less components than those shown, or may combine some of the components, or have a different arrangement of components.
In one embodiment, a computer device is provided comprising a memory and a processor, the memory having stored therein a computer program, the processor when executing the computer program performing the steps of:
receiving a search statement input by a user; inputting the search statement into a functional identification network model with the training completed; processing the search statement according to the function identification network model, and determining the confidence coefficient of each function label; the function tag comprises function information and platform information; and determining a display sequence according to the confidence coefficient of each function label, and displaying a recommended function entry corresponding to each function label based on the display sequence.
In one embodiment, the processor when executing the computer program further performs the steps of:
Under the condition that the current historical behavior data of the user reaches a preset condition, determining the priority of each function label based on the current historical behavior data of the user; determining the display sequence of each function label based on the priority of each function label and the confidence of each function label, and displaying the recommended function entry corresponding to each function label based on the display sequence of each function label.
In one embodiment, the processor when executing the computer program further performs the steps of:
updating the training sample set under the condition that an updating instruction is detected; the training sample set consists of a plurality of search statement training samples marked with function labels; acquiring a current training sample set; inputting the current training sample set to an initial function identification network model, and training the initial function identification network model; under the condition that training is carried out until the initial function identification network model meets a preset loss condition, determining that the initial function identification network model training is completed; and taking the initial functional identification network model after training as the functional identification network model after training.
In one embodiment, the processor when executing the computer program further performs the steps of:
under the condition that the update instruction is detected to be a new first function label instruction, acquiring a search statement training sample marked with the first function label, adding the search statement training sample of the first function label into the training sample set, and updating the training sample set.
In one embodiment, the processor when executing the computer program further performs the steps of:
and deleting the search statement training sample marked with the second function label in the training sample set under the condition that the updating instruction is detected to be the second function label deleting instruction, and updating the training sample set.
In one embodiment, the processor when executing the computer program further performs the steps of:
under the condition that the preset moment is reached, acquiring a current training sample set; the preset time is a period arrival time or a preset time.
In one embodiment, the processor when executing the computer program further performs the steps of:
releasing the trained function recognition network model to a search interface; and calling the functional identification network model which is completed through training through a search interface, and inputting the search statement into the functional identification network model.
In one embodiment, the processor when executing the computer program further performs the steps of:
the query function identifies whether the training of the network model is completed; and under the condition that the training of the functional identification network model is completed, inputting the search statement into the functional identification network model with the completed training.
In one embodiment, a computer readable storage medium is provided having a computer program stored thereon, which when executed by a processor, performs the steps of:
receiving a search statement input by a user; inputting the search statement into a functional identification network model with the training completed; processing the search statement according to the function identification network model, and determining the confidence coefficient of each function label; the function tag comprises function information and platform information; and determining a display sequence according to the confidence coefficient of each function label, and displaying a recommended function entry corresponding to each function label based on the display sequence.
In one embodiment, the computer program when executed by the processor further performs the steps of:
under the condition that the current historical behavior data of the user reaches a preset condition, determining the priority of each function label based on the current historical behavior data of the user; determining the display sequence of each function label based on the priority of each function label and the confidence of each function label, and displaying the recommended function entry corresponding to each function label based on the display sequence of each function label.
In one embodiment, the computer program when executed by the processor further performs the steps of:
updating the training sample set under the condition that an updating instruction is detected; the training sample set consists of a plurality of search statement training samples marked with function labels; acquiring a current training sample set; inputting the current training sample set to an initial function identification network model, and training the initial function identification network model; under the condition that training is carried out until the initial function identification network model meets a preset loss condition, determining that the initial function identification network model training is completed; and taking the initial functional identification network model after training as the functional identification network model after training.
In one embodiment, the computer program when executed by the processor further performs the steps of:
under the condition that the update instruction is detected to be a new first function label instruction, acquiring a search statement training sample marked with the first function label, adding the search statement training sample of the first function label into the training sample set, and updating the training sample set.
In one embodiment, the computer program when executed by the processor further performs the steps of:
And deleting the search statement training sample marked with the second function label in the training sample set under the condition that the updating instruction is detected to be the second function label deleting instruction, and updating the training sample set.
In one embodiment, the computer program when executed by the processor further performs the steps of:
under the condition that the preset moment is reached, acquiring a current training sample set; the preset time is a period arrival time or a preset time.
In one embodiment, the computer program when executed by the processor further performs the steps of:
releasing the trained function recognition network model to a search interface; and calling the functional identification network model which is completed through training through a search interface, and inputting the search statement into the functional identification network model.
In one embodiment, the computer program when executed by the processor further performs the steps of:
the query function identifies whether the training of the network model is completed; and under the condition that the training of the functional identification network model is completed, inputting the search statement into the functional identification network model with the completed training.
In one embodiment, a computer program product is provided comprising a computer program which, when executed by a processor, performs the steps of:
Receiving a search statement input by a user; inputting the search statement into a functional identification network model with the training completed; processing the search statement according to the function identification network model, and determining the confidence coefficient of each function label; the function tag comprises function information and platform information; and determining a display sequence according to the confidence coefficient of each function label, and displaying a recommended function entry corresponding to each function label based on the display sequence.
In one embodiment, the computer program when executed by the processor further performs the steps of:
under the condition that the current historical behavior data of the user reaches a preset condition, determining the priority of each function label based on the current historical behavior data of the user; determining the display sequence of each function label based on the priority of each function label and the confidence of each function label, and displaying the recommended function entry corresponding to each function label based on the display sequence of each function label.
In one embodiment, the computer program when executed by the processor further performs the steps of:
updating the training sample set under the condition that an updating instruction is detected; the training sample set consists of a plurality of search statement training samples marked with function labels; acquiring a current training sample set; inputting the current training sample set to an initial function identification network model, and training the initial function identification network model; under the condition that training is carried out until the initial function identification network model meets a preset loss condition, determining that the initial function identification network model training is completed; and taking the initial functional identification network model after training as the functional identification network model after training.
In one embodiment, the computer program when executed by the processor further performs the steps of:
under the condition that the update instruction is detected to be a new first function label instruction, acquiring a search statement training sample marked with the first function label, adding the search statement training sample of the first function label into the training sample set, and updating the training sample set.
In one embodiment, the computer program when executed by the processor further performs the steps of:
and deleting the search statement training sample marked with the second function label in the training sample set under the condition that the updating instruction is detected to be the second function label deleting instruction, and updating the training sample set.
In one embodiment, the computer program when executed by the processor further performs the steps of:
under the condition that the preset moment is reached, acquiring a current training sample set; the preset time is a period arrival time or a preset time.
In one embodiment, the computer program when executed by the processor further performs the steps of:
releasing the trained function recognition network model to a search interface; and calling the functional identification network model which is completed through training through a search interface, and inputting the search statement into the functional identification network model.
In one embodiment, the computer program when executed by the processor further performs the steps of:
the query function identifies whether the training of the network model is completed; and under the condition that the training of the functional identification network model is completed, inputting the search statement into the functional identification network model with the completed training.
It should be noted that, user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, presented data, etc.) referred to in the present application are information and data authorized by the user or sufficiently authorized by each party.
Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, database, or other medium used in the various embodiments provided herein may include at least one of non-volatile and volatile memory. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, high density embedded nonvolatile Memory, resistive random access Memory (ReRAM), magnetic random access Memory (Magnetoresistive Random Access Memory, MRAM), ferroelectric Memory (Ferroelectric Random Access Memory, FRAM), phase change Memory (Phase Change Memory, PCM), graphene Memory, and the like. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory, and the like. By way of illustration, and not limitation, RAM can be in the form of a variety of forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM), and the like. The databases referred to in the various embodiments provided herein may include at least one of relational databases and non-relational databases. The non-relational database may include, but is not limited to, a blockchain-based distributed database, and the like. The processors referred to in the embodiments provided herein may be general purpose processors, central processing units, graphics processors, digital signal processors, programmable logic units, quantum computing-based data processing logic units, etc., without being limited thereto.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The above examples only represent a few embodiments of the present application, which are described in more detail and are not to be construed as limiting the scope of the present application. It should be noted that it would be apparent to those skilled in the art that various modifications and improvements could be made without departing from the spirit of the present application, which would be within the scope of the present application. Accordingly, the scope of protection of the present application shall be subject to the appended claims.

Claims (12)

1. A search method, wherein the method is applied to an application integrating multi-platform functionality, the method comprising:
receiving a search statement input by a user;
inputting the search statement into a functional identification network model with the training completed;
processing the search statement according to the function identification network model, and determining the confidence coefficient of each function label; the function tag comprises function information and platform information;
And determining a display sequence according to the confidence coefficient of each function label, and displaying a recommended function entry corresponding to each function label based on the display sequence.
2. The method of claim 1, wherein determining a display order according to the confidence level of each of the function labels, and displaying the recommended function entry corresponding to each of the function labels based on the display order, comprises:
under the condition that the current historical behavior data of the user reaches a preset condition, determining the priority of each function label based on the current historical behavior data of the user;
determining the display sequence of each function label based on the priority of each function label and the confidence of each function label, and displaying the recommended function entry corresponding to each function label based on the display sequence of each function label.
3. The method according to claim 1, wherein the method further comprises:
updating the training sample set under the condition that an updating instruction is detected; the training sample set consists of a plurality of search statement training samples marked with function labels;
acquiring a current training sample set;
Inputting the current training sample set to an initial function identification network model, and training the initial function identification network model;
under the condition that training is carried out until the initial function identification network model meets a preset loss condition, determining that the initial function identification network model training is completed;
and taking the initial functional identification network model after training as the functional identification network model after training.
4. A method according to claim 3, wherein in the event that an update instruction is detected, updating the training sample set comprises:
under the condition that the update instruction is detected to be a new first function label instruction, acquiring a search statement training sample marked with the first function label, adding the search statement training sample of the first function label into the training sample set, and updating the training sample set.
5. A method according to claim 3, wherein in the event that an update instruction is detected, updating the training sample set comprises:
and deleting the search statement training sample marked with the second function label in the training sample set under the condition that the updating instruction is detected to be the second function label deleting instruction, and updating the training sample set.
6. A method according to claim 3, wherein said obtaining a current training sample set comprises:
under the condition that the preset moment is reached, acquiring a current training sample set; the preset time is a period arrival time or a preset time.
7. A method according to claim 1 or 3, wherein after taking the trained initial function recognition network model as the trained function recognition network model, the method further comprises:
releasing the trained function recognition network model to a search interface;
the step of inputting the search statement into the trained function recognition network model comprises the following steps:
and calling the functional identification network model which is completed through training through a search interface, and inputting the search statement into the functional identification network model.
8. The method of claim 1, wherein said inputting the search statement into the trained functional recognition network model comprises:
the query function identifies whether the training of the network model is completed;
and under the condition that the training of the functional identification network model is completed, inputting the search statement into the functional identification network model with the completed training.
9. A search apparatus for use in an application integrating multi-platform functionality, the apparatus comprising:
the receiving module is used for receiving search sentences input by a user;
the input module is used for inputting the search statement into the functional identification network model after training;
the determining module is used for processing the search statement according to the function identification network model and determining the confidence coefficient of each function label; the function tag comprises function information and platform information;
and the display module is used for determining a display sequence according to the confidence coefficient of each function label and displaying recommended function entries corresponding to each function label based on the display sequence.
10. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any one of claims 1 to 8 when the computer program is executed.
11. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 8.
12. A computer program product comprising a computer program, characterized in that the computer program, when executed by a processor, implements the steps of the method of any one of claims 1 to 8.
CN202310383445.1A 2023-04-11 2023-04-11 Searching method, searching device, computer equipment and storage medium Pending CN116383508A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310383445.1A CN116383508A (en) 2023-04-11 2023-04-11 Searching method, searching device, computer equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310383445.1A CN116383508A (en) 2023-04-11 2023-04-11 Searching method, searching device, computer equipment and storage medium

Publications (1)

Publication Number Publication Date
CN116383508A true CN116383508A (en) 2023-07-04

Family

ID=86978494

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310383445.1A Pending CN116383508A (en) 2023-04-11 2023-04-11 Searching method, searching device, computer equipment and storage medium

Country Status (1)

Country Link
CN (1) CN116383508A (en)

Similar Documents

Publication Publication Date Title
CN108287864B (en) Interest group dividing method, device, medium and computing equipment
CN107657048A (en) user identification method and device
CN110119473A (en) A kind of construction method and device of file destination knowledge mapping
CN111159431A (en) Knowledge graph-based information visualization method, device, equipment and storage medium
CN113051409A (en) Multi-dimensional information calculation-based business opportunity recommendation system and storage medium for industry of know-produce
CN112381236A (en) Data processing method, device, equipment and storage medium for federal transfer learning
CN116957006A (en) Training method, device, equipment, medium and program product of prediction model
CN116089595A (en) Data processing pushing method, device and medium based on scientific and technological achievements
CN116383508A (en) Searching method, searching device, computer equipment and storage medium
CN115409111A (en) Training method of named entity recognition model and named entity recognition method
CN114780745A (en) Method and device for constructing knowledge system, electronic equipment and storage medium
CN110837596B (en) Intelligent recommendation method and device, computer equipment and storage medium
CN116757216B (en) Small sample entity identification method and device based on cluster description and computer equipment
CN114168787A (en) Music recommendation method and device, computer equipment and storage medium
CN114118059A (en) Sample statement processing method and device, computer equipment and storage medium
CN117033451A (en) Searching method, searching device, computer equipment and storage medium
CN116910604A (en) User classification method, apparatus, computer device, storage medium, and program product
CN116244521A (en) Service acquisition method, device, equipment and storage medium of software platform
CN116301786A (en) Auxiliary encoding method, device, computer equipment and storage medium
CN117931844A (en) Policy generation method, policy generation device, computer equipment and storage medium
CN115374294A (en) Content recommendation method, apparatus, computer device, storage medium, and program product
CN116881544A (en) Financial product information pushing method, device, computer equipment and storage medium
CN117788078A (en) Quality evaluation method, quality evaluation device, computer device, and storage medium
CN117033766A (en) Service processing behavior prediction method, device, computer equipment and storage medium
CN117708428A (en) Recommendation information prediction method and device and electronic equipment

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination