CN116737762B

CN116737762B - Structured query statement generation method, device and computer readable medium

Info

Publication number: CN116737762B
Application number: CN202310987876.9A
Authority: CN
Inventors: 陈家耀; 赖林华; 陈俊豪; 吕亮
Original assignee: Beijing Hengshi Technology Co ltd
Current assignee: Zhongguancun Technology Leasing Co ltd
Priority date: 2023-08-08
Filing date: 2023-08-08
Publication date: 2023-10-27
Anticipated expiration: 2043-08-08
Also published as: CN116737762A

Abstract

Embodiments of the present disclosure disclose a structured query statement generation method, apparatus, and computer-readable medium. One embodiment of the method comprises the following steps: acquiring query text information and query index definition information corresponding to the query text information; generating candidate index information according to the query text information in response to determining that the preset index name table does not contain the target index name; generating a model, query text information and candidate index information according to the index query information, and generating index query information; generating a model and index business logic information through the structured query statement, and generating a structured query statement corresponding to the index query information; and carrying out data query on the target database according to the structured query statement to generate query result information. The embodiment realizes automatic writing of the structured sentences and improves writing efficiency.

Description

Structured query statement generation method, device and computer readable medium

Technical Field

Embodiments of the present disclosure relate to the field of computer technology, and in particular, to a method, an apparatus, and a computer readable medium for generating a structured query statement.

Background

With the popularization of data electronization, databases are widely used as storage media for (electronic) data due to their strong data storage capacity. Currently, when processing data stored in a database (e.g., data query), the following methods are generally adopted: and manually writing a structured query statement to realize the processing of the data in the database.

However, the inventors found that when the above manner is adopted, there are often the following technical problems:

firstly, the writing of the structured query statement often requires a writer to have certain writing capability, and meanwhile, aiming at a large number of different data query demands, a mode of manually writing the structured query statement is adopted, so that the writing efficiency is low;

secondly, when index names corresponding to the query text information are inaccurate, the generated structured query statement is invalid in data query, so that data query resources are wasted;

third, when a structured query sentence is generated, as the feature extraction depth deepens, there may be a problem of feature loss, thereby affecting the accuracy of the generated structured query sentence.

The above information disclosed in this background section is only for enhancement of understanding of the background of the inventive concept and, therefore, may contain information that does not form the prior art that is already known to those of ordinary skill in the art in this country.

Disclosure of Invention

The disclosure is in part intended to introduce concepts in a simplified form that are further described below in the detailed description. The disclosure is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter.

Some embodiments of the present disclosure propose a structured query statement generation method, apparatus, and computer-readable medium to address one or more of the technical problems mentioned in the background section above.

In a first aspect, some embodiments of the present disclosure provide a method for generating a structured query statement, the method comprising: acquiring query text information input by a target user and query index definition information corresponding to the query text information, wherein the query index definition information comprises: target index name and index business logic information; generating candidate index information according to the query text information in response to determining that the target index name is not contained in a preset index name table; generating an index query information according to a pre-trained index query information generation model, the query text information and the candidate index information, wherein the index query information is structured information; generating a structured query statement corresponding to the index query information through a pre-trained structured query statement generation model and the index business logic information; and carrying out data query on the target database according to the structured query statement so as to generate query result information.

In a second aspect, some embodiments of the present disclosure provide a structured query statement generation apparatus, the apparatus comprising: the acquisition unit is configured to acquire query text information input by a target user and query index definition information corresponding to the query text information, wherein the query index definition information comprises: target index name and index business logic information; the first generation unit is configured to generate candidate index information according to the query text information in response to determining that the target index name is not contained in the index name table which is set in advance; the second generation unit is configured to generate an index query information according to a pre-trained index query information generation model, the query text information and the candidate index information, wherein the index query information is structured information; the third generating unit is configured to generate a structured query sentence corresponding to the index query information through a pre-trained structured query sentence generating model and the index business logic information; the data query unit is configured to perform data query on the target database according to the structured query statement so as to generate query result information.

In a third aspect, some embodiments of the present disclosure provide an electronic device comprising: one or more processors; a storage device having one or more programs stored thereon, which when executed by one or more processors causes the one or more processors to implement the method described in any of the implementations of the first aspect above.

In a fourth aspect, some embodiments of the present disclosure provide a computer readable medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the method described in any of the implementations of the first aspect above.

The above embodiments of the present disclosure have the following advantages: by the method for generating the structured query statement in some embodiments of the present disclosure, automatic writing of the structured statement is achieved, and writing efficiency is improved. Specifically, the reason for the inefficiency of writing is that: the writing of the structured query statement often requires personnel to have certain writing capability, and meanwhile, the writing efficiency is low by adopting a mode of manually writing the structured query statement according to a large number of different data query demands. Based on this, in the method for generating a structured query sentence according to some embodiments of the present disclosure, first, query text information input by a target user and query index definition information corresponding to the query text information are obtained, where the query index definition information includes: target index name and index business logic information. By acquiring the query text information input by the target user and the query index definition information corresponding to the query text information, the name of the target index of the query and the corresponding index service logic information can be determined. And then, generating the candidate index information according to the query text information in response to determining that the target index name is not contained in a preset index name table. In practice, there are some scenarios, that is, the target index corresponding to the query text information is not directly contained in the index name table, and when the index name is absent in the structured query statement, the data query amount may be increased, resulting in waste of computing resources. At the same time, there is also failure in execution of the structured query statement, and therefore, when the target index name is not included in the index name table, candidate index information needs to be generated. And generating index query information according to a pre-trained index query information generation model, the query text information and the candidate index information, wherein the index query information is structured information. In practice, index query information is generated, so that the query requirement can be expressed more directly, redundant and fuzzy expression is avoided, the query process is optimized, and the query efficiency and performance are improved. In addition, the structured query statement corresponding to the index query information is generated through a pre-trained structured query statement generation model and the index business logic information. In practice, a pre-trained structured query statement generation model is utilized to generate the structured query statement corresponding to the index query information. Thus, automatic and efficient structured query statement generation for query text information is achieved. And finally, carrying out data query on the target database according to the structured query statement so as to generate query result information. By the method, automatic writing of the structured sentences is realized, writing efficiency is improved, and timeliness of data query is also improved on the side face.

Drawings

The above and other features, advantages, and aspects of embodiments of the present disclosure will become more apparent by reference to the following detailed description when taken in conjunction with the accompanying drawings. The same or similar reference numbers will be used throughout the drawings to refer to the same or like elements. It should be understood that the figures are schematic and that elements and components are not necessarily drawn to scale.

FIG. 1 is a flow chart of some embodiments of a structured query statement generation method according to the present disclosure;

FIG. 2 is a schematic diagram of some embodiments of a structured query statement generation device according to the present disclosure;

fig. 3 is a schematic structural diagram of an electronic device suitable for use in implementing some embodiments of the present disclosure.

Detailed Description

Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While certain embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete. It should be understood that the drawings and embodiments of the present disclosure are for illustration purposes only and are not intended to limit the scope of the present disclosure.

It should be noted that, for convenience of description, only the portions related to the present invention are shown in the drawings. Embodiments of the present disclosure and features of embodiments may be combined with each other without conflict.

It should be noted that the terms "first," "second," and the like in this disclosure are merely used to distinguish between different devices, modules, or units and are not used to define an order or interdependence of functions performed by the devices, modules, or units.

It should be noted that references to "one", "a plurality" and "a plurality" in this disclosure are intended to be illustrative rather than limiting, and those of ordinary skill in the art will appreciate that "one or more" is intended to be understood as "one or more" unless the context clearly indicates otherwise.

The names of messages or information interacted between the various devices in the embodiments of the present disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

The present disclosure will be described in detail below with reference to the accompanying drawings in conjunction with embodiments.

Referring to FIG. 1, a flow 100 of some embodiments of a structured query statement generation method according to the present disclosure is shown. The method for generating the structured query statement comprises the following steps:

Step 101, acquiring query text information input by a target user and query index definition information corresponding to the query text information.

In some embodiments, an execution subject (e.g., a computing device) of the structured query statement generation method may obtain query text information input by a target user and query index definition information corresponding to the query text information. Wherein, the query index definition information includes: target index name and index business logic information. The target user may be a user who needs to convert the query text information into a structured query term. In practice, the target user may be a user having access to the target database. The target database may be a database to be queried for data. The query text information may be natural language for making data queries. For example, the query text information may be "average sales of query A channel or B channel". The query index definition information may be predefined related information for the target index name. The target index name may be an index name corresponding to the above-mentioned query text information. For example, the target index name may be "average sales". The index business logic information may characterize the index calculation logic corresponding to the target index name. In practice, the executing entity may obtain the query text information from a query text information input interface.

Step 102, in response to determining that the preset index name table does not contain the target index name, candidate index information is generated according to the query text information.

In some embodiments, in response to determining that the target index name is not included in the index name table set in advance, the execution body may generate the candidate index information according to the query text information. The index name table may be a data table of each index name included in at least one data table in the target database. The target database may be a database to be queried for data. Wherein the candidate index information may characterize temporary index names to be used for data retrieval. For example, when the "average sales" is not included in the index name table, "average" may be used as the candidate index information.

Optionally, the index name table includes: index name set. Wherein the index name set may be each index name included in at least one data table in the target database.

In some optional implementations of some embodiments, the executing entity may generate candidate index information according to the query text information, and the method may include the following steps:

And firstly, performing word segmentation processing on the query text information to obtain a word set.

In practice, the execution subject may perform word segmentation through the jieba library to generate a word set. Wherein the jieba library is a Chinese word stock of Python.

As an example, the query text information may be "average sales of query a channel or B channel". Thus, the term set may be [ query, A-channel, or B-channel, average sales ].

Second, for each word in the word set, the following processing steps are performed:

sub-step 1: and carrying out word shielding on the words included in the query text information to obtain shielded query text information.

Wherein the execution subject may block the word by a blocking symbol. The occlusion symbol may be a symbol for word occlusion. For example, the term may be "query", the MASK may be "[ MASK ]", and the resulting blocked query text information may be "[ MASK ] average sales of a channel or B channel.

Sub-step 2: and carrying out text coding on the shielded query text information through a coding model included in the pre-trained part-of-speech determination model so as to generate coded query text information.

In practice, the part-of-speech determination model may be a model for determining part-of-speech information corresponding to a word. The coding model may be a model for information coding the query text information after occlusion. In particular, the coding model may be a model based on a transducer structure.

Sub-step 3: and inputting the coded query text information into a decoding model included in the part-of-speech determination model to generate decoded query text information.

In practice, the decoding model is a model for information decoding of the encoded query text information. The coding model can also be a model based on a transducer structure. The model structures of the coding model and the decoding model are symmetrical. The decoded query text information may be a decoded feature vector of the information.

Sub-step 4: and marking the parts of speech of the decoded query text information through a pre-trained part of speech marking model so as to generate part of speech information corresponding to the word.

In practice, the part-of-speech tagging model is a model for determining the part of speech to which a word corresponds to generate part-of-speech information. Specifically, the part-of-speech tagging model may be a Seq2Seq model.

As an example, a word may be a "query" and the corresponding part-of-speech information may be "v".

And thirdly, screening out words with the corresponding part-of-speech information as target part-of-speech information from the word set, and taking the words as candidate words to obtain a candidate word set.

Wherein the target part-of-speech information may be an "n" token noun.

As an example, the term set may be [ query average sales of a channel or B channel ]. Wherein, the part-of-speech information corresponding to the "query" may be "v". The "part-of-speech information" corresponding to "channel A" may be "adj". The "or" corresponding part-of-speech information may be "n". The "part-of-speech information" corresponding to "B channel" may be "adj". The part-of-speech information corresponding to the "average sales" may be "n". The resulting set of candidate words may be [ or average sales ].

And a fourth step of determining, for each candidate word in the candidate word set, index name similarity between the candidate word and each index name in the index name set as a first index name similarity, and obtaining a first index name similarity group.

Wherein the first index name similarity characterizes a similarity between the candidate word and the index name.

In practice, first, the execution subject may perform word vector encoding on the candidate word to generate a candidate word vector. The execution body may then perform word vector encoding on the index names to generate index name vectors. Then, the execution body may determine cosine similarity between the candidate word vector and the index name vector as the first index name similarity.

And fifthly, screening candidate words meeting the first screening condition from the candidate word set to serve as target words.

Wherein, the first screening condition is: and the first index name similarity group corresponding to the candidate word has the first index name similarity which is larger than the first preset index name similarity and smaller than the second preset index name similarity.

In practice, the first preset index name similarity may be 0.2, and the second preset index name similarity may be 0.95.

As an example, the candidate set of words may be [ or average sales ]. Wherein the first index name similarity group corresponding to or may be [0 0 0 0 0]. The first index name similarity group corresponding to the "average sales" may be [ 0.1.6.0.0.1 ]. Wherein 0.2 < 0.6 < 0.95, and thus the target word is "average sales".

And sixthly, determining index calculation information corresponding to the target word.

In practice, the index calculation information characterizes a calculation function corresponding to the target word. Specifically, first, the execution subject may determine word sense information corresponding to the target word. Then, the execution subject may screen out a calculation function matching the corresponding calculation function description information and the word sense information from a calculation function library according to the word sense information, as the index calculation information. The computing function description information is used for describing the function of the computing function. In practice, the execution body may screen the computing function matching the corresponding computing function description information and the word sense information from the computing function library as the index computing information by determining that the word sense information is semantically similar to the computing function description information.

As an example, the target word may be an "average sales," and the corresponding word sense information may be an "average value for determining sales. The calculation function description information corresponding to the "calculation function average ()" may be "average value determining at least one numerical value", and thus, the index calculation information may be "average ()".

Seventh, candidate index information is generated according to the index calculation information and the target word.

The execution subject may use a part of words in the target word as function parameters of a calculation function corresponding to the index calculation information, so as to obtain the candidate index information. For example, the target word may be "average sales", the index calculation information may be "average ()", and the candidate index information may be "average".

The content of the foregoing "in some optional implementations of some embodiments" is taken as an invention point of the disclosure, which solves the second technical problem mentioned in the background art, namely "when the index name corresponding to the query text information is inaccurate, the generated structured query statement will cause query failure when the data query is performed, so that the data query resource is wasted. In practice, since the query text information is natural language, that is, the index names corresponding to the query text information may be spoken language expressions, a structured query sentence is constructed based on the index names corresponding to the query text information, and when data query is performed according to the constructed structured query sentence, query failure is caused, so that waste of data query resources (such as computing resources and database retrieval resources) is caused. Based on the above, the present disclosure firstly performs word segmentation processing on the above-mentioned query text information to obtain a word set, and in practice, the query text information often includes corresponding index names, so that each possible index name with a word as granularity can be obtained by a word segmentation method. Next, for each word in the above-described word set, the following processing steps are performed: word shielding is carried out on the words included in the query text information, and the query text information after shielding is obtained; performing text coding on the blocked query text information through a coding model included in a pre-trained part-of-speech determination model to generate coded query text information; and inputting the coded query text information into a decoding model included in the part-of-speech determination model to generate decoded query text information. Because the words in the word set have word dependence with the words before and after in the query text information, the robustness of the obtained features can be improved by carrying out feature processing in a word shielding mode. Further, part-of-speech tagging is carried out on the decoded query text information through a pre-trained part-of-speech tagging model so as to generate part-of-speech information corresponding to the word; and screening out the words with the corresponding part-of-speech information as the target part-of-speech information from the word set, and taking the words as candidate words to obtain a candidate word set. In practice, since the query text information is natural language, the query text information satisfies basic sentence structure, such as a main predicate or a main system table, that is, the query text information is composed of words with various parts of speech. The part of speech of the index name is often a noun, and the part of speech of the words in the word set can be accurately marked through the part of speech marking model. Finally, for each candidate word in the candidate word set, determining the index name similarity of each index name in the candidate word set and the index name set as a first index name similarity, and obtaining a first index name similarity group; screening candidate words meeting a first screening condition from the candidate word set to serve as target words, wherein the first screening condition is as follows: the first index name similarity group corresponding to the candidate word has first index name similarity which is larger than the first preset index name similarity and smaller than the second preset index name similarity; determining index calculation information corresponding to the target word; and generating the candidate index information according to the index calculation information and the target word. Candidate index information is obtained by calculating the similarity, so that when index names corresponding to the query text information are not contained in the index name table, the index names corresponding to the query text information are replaced, the performability of the subsequently generated structured query statement is ensured, the success probability of executing the structured query statement is improved, and the problem that data query resources are wasted due to failure of the structured query statement is solved.

In some optional implementations of some embodiments, after the executing entity generates the candidate index information according to the query text information in response to determining that the target index name is not included in the index name table set in advance, the method further includes:

and determining the index name similarity of the target index name and each index name in the index name set as a second index name similarity to obtain a second index name similarity group.

Wherein the second index name similarity characterizes a similarity between the target index name and the index name. In practice, first, the execution subject may perform word vector encoding on the target index name to generate a target index name vector. The execution body may then perform word vector encoding on the index names to generate index name vectors. Then, the execution body may determine cosine similarity between the target index name vector and the index name vector as the second index name similarity.

And a second step of determining the target index name as the candidate index information in response to determining that there is a second index name similarity satisfying a second screening condition in the second index name similarity group.

Wherein, the second screening condition is: the second index name similarity is greater than the second preset index name similarity.

In practice, the second preset index name similarity may be 0.95.

As an example, the target index name may be "average sales". The second index name similarity group for which "average sales" corresponds may be [ 0.1.99.0.0.1 ]. Wherein 0.95 < 0.99, and thus the candidate index information is "average sales".

And step 103, generating a model, query text information and candidate index information according to the pre-trained index query information, and generating index query information.

In some embodiments, the executing entity may generate the index query information from a pre-trained index query information generation model, the query text information, and the candidate index information. Wherein the index query information is structured information. Wherein the index query information generation model may be a model for generating structured, index query information. In practice, the index query generation model may be a GPT model. Wherein the query text information and the candidate index information may be model inputs of an index query information generation model. The index query information may be a model output of an index query information generation model.

As an example, the query text information may be "average sales of query a channel or B channel". The candidate index information may be "average". The index query information may be:

{

index name: average (sales);

filtration conditions: channel= "a" or channel= "B";

}。

optionally, the index query information generation model includes: the method comprises a first text feature extraction model, a second text feature extraction model and a fusion feature extraction model. The first text feature extraction model is used for extracting features of the query text information. The second text feature extraction model is used for extracting features of the candidate index information. The fusion feature extraction model is used for carrying out feature fusion and further feature extraction on the first text feature extracted by the first text feature extraction model and the second text feature extracted by the second text feature extraction model.

In some optional implementations of some embodiments, the executing entity may generate the index query information according to a pre-trained index query information generating model, the query text information, and the candidate index information, and the generating may include the following steps:

And firstly, extracting the characteristics of the query text information through the first text characteristic extraction model so as to generate first text characteristics.

Wherein the first text feature extraction model comprises: a text conversion model, a semantic feature extraction model selector and a semantic feature extraction model. The semantic feature extraction model includes: a first semantic feature extraction model and a second semantic feature extraction model. The text conversion model is used for converting the query text information into corresponding vector representations. Wherein the text conversion model includes K hidden layers. Wherein the K hidden layers are connected in series. The semantic feature extraction model selector is operable to determine whether to use the output of the text conversion model as an input to the first semantic feature extraction model or as an input to the second semantic feature extraction model. In practice, the semantic feature extraction model selector may be a fully connected layer. Specifically, the semantic feature extraction model selector may output a feature vector of 1×2. Wherein the 1 st dimension of the 1 x 2 feature vector characterizes the selection of the output of the text conversion model as the input of the first semantic feature extraction model. The 2 nd dimension representation of the feature vector of 1 x 2 selects the output of the text conversion model as the input of the second semantic feature extraction model. For example, a 1 x 2 feature vector may be [0.98,0], i.e., characterizing the choice of choosing the output of the text conversion model as the input of the first semantic feature extraction model. The first semantic feature extraction model may include K serially connected convolutional layers, i.e., the first semantic feature extraction model adopts a convolutional neural network as a basic network structure. The second semantic feature extraction model may employ a recurrent neural network as the underlying network structure.

As an example, first, the above-described execution subject may input query text information into a text conversion model to generate a converted feature vector. The execution body may then input the converted feature vector to a semantic feature extraction model selector to generate a 1×2 feature vector. Then, the execution body may select whether to use the first semantic feature extraction model or the second semantic feature extraction model according to the feature vector of 1×2, and perform feature extraction on the converted feature vector to generate the first text feature.

And secondly, extracting the characteristics of the candidate index information through the second text characteristic extraction model so as to generate second text characteristics.

In practice, the second text feature extraction model may share the text conversion model and the first semantic feature extraction model comprised by the first text feature extraction model.

And thirdly, inputting the first text feature and the second text feature into the fusion feature extraction model to generate the index query information.

The fusion feature extraction model comprises the following steps: feature stitching layer and GPT model. The execution main body can perform feature splicing on the first text feature and the second text feature through the feature splicing layer to obtain spliced features. Then, the execution body may input the spliced features into the GPT model to obtain index query information.

And 104, generating a structured query statement corresponding to the index query information through a pre-trained structured query statement generation model and the index business logic information.

In some embodiments, the executing entity may generate the structured query statement corresponding to the index query information through a pre-trained structured query statement generation model and the index business logic information. The structured query statement generation model is used for converting the input index query information and index business logic information into structured query statements corresponding to the index query information. In practice, the structured query term generation model may be implemented using a hard-coded or structured query term generation model. The structured query statement may be an SQL (Structured Query Language ) statement for data retrieval of the target database, among others.

AS an example, the index service logic information may be "SELECT product_id, AVG (quality) AS average_ sales FROM sales WHERE product _id". The index query information may be:

{

index name: average (sales);

filtration conditions: channel= "a" or channel= "B";

}。

The structured query statement may be "SELECT average (sales) FROM post-table data table WHERE channel=" a "OR channel=" B "".

Optionally, the structured query statement generation model includes: an information feature extraction network model and an information transformation network model.

In some optional implementations of some embodiments, the generating, by the execution body, the structured query statement corresponding to the index query information through the pre-trained structured query statement generation model and the index business logic information may include the following steps:

and firstly, extracting information features of the index business logic information and the index query information through the information feature extraction network model to generate information extracted features.

Wherein the information feature extraction network model may multiplex the first text feature extraction model.

And secondly, carrying out pooling operation on the information extracted features to generate pooled features.

In practice, the execution body may perform an average pooling operation on the information extracted features to generate pooled features.

And thirdly, inputting the pooled features into the information conversion network model to generate the structured query statement.

In practice, the information conversion network model may multiplex the GPT model included in the fusion feature extraction model.

The content of "in some alternative implementations of some embodiments" in step 103 and step 104, as an invention point of the present disclosure, solves the technical problem three mentioned in the background art, namely, "when a structured query statement is generated, as the feature extraction depth is deepened, there may be a problem that the feature is lost, thereby affecting the accuracy of the generated structured query statement. In practice, it is often necessary to combine query text information with candidate index information in generating index query information. Since the query text information is the content to be queried described in the form of natural language. When corresponding feature extraction is carried out, the problem of feature loss is easily caused along with the deepening of the feature extraction depth when text information is searched for too long. Based on this, the present disclosure designs a first text feature extraction model. First, the text information of the query in text form is converted into computable features by a text conversion model comprised by the first text feature extraction model. Then, considering that the deeper convolutional neural network is very easy to cause feature loss when aiming at longer inquiry text information, and the cyclic neural network is adopted for shorter inquiry text information, the data calculation amount is increased. Thus, the present disclosure designs a semantic feature extraction model selector to automatically determine whether to employ a first semantic feature extraction model or a second semantic feature extraction model to further feature extract the output of the text conversion model. Thereby improving the extraction efficiency and accuracy of the features. Next, considering that the candidate index information is actually an alias of the target index name, that is, a length is short, the text conversion model and the first semantic feature extraction model included in the first text feature extraction model may be shared. To reduce the training costs for model training. In addition, in the process of generating the structured query term, the feature extraction for the index business logic information and the index query information is similar to the feature extraction task corresponding to the query text information, so that the first text feature extraction model can be further multiplexed as the information feature extraction network model. Finally, considering that the generation of the structured query statement and the index query information are similar generation tasks, the GPT model included in the fusion feature extraction model can be shared as an information conversion network model. By the method, on the premise of ensuring the accuracy of the generated structured query statement, the volume of the model is reduced, the parameters of the model are reduced, the computational complexity is reduced, and meanwhile, the hardware resource cost and the time cost consumed by training the model are reduced on the side face.

And 105, carrying out data query on the target database according to the structured query statement to generate query result information.

In some embodiments, the execution body may perform a data query on the target database according to the structured query statement to generate query result information. The query result information characterizes a query result corresponding to the query text information in the target database. The execution body can perform data query by executing the structured query statement in the target database to obtain query result information. In practice, the execution body may execute the structured query statement to implement data retrieval on the target database, so as to obtain the query result information.

In some optional implementations of some embodiments, the executing body performs a data query on the target database according to the structured query statement to generate query result information, and may include the following steps:

the first step is to analyze the above structured query sentence to determine the information set of the data table to be searched.

The data table information to be searched characterizes a data table to be subjected to data searching in the target database.

For example, the data table information set to be searched may be a data table corresponding to the "a channel" and a data table corresponding to the "B channel".

And secondly, carrying out table connecting operation on the data table corresponding to the data table information set to be searched in the target database to obtain a data table after table connecting.

The data table after the table connection can be a full data table of each data table corresponding to the data table information set to be searched.

As an example, the executing body may perform a table linking operation on the data table of the channel a and the data table of the channel B, to obtain a data table after the table linking.

And thirdly, executing the structured query statement, and carrying out data query on the linked data table to generate the query result information.

As an example, the structured query statement may be "SELECT average (sales) FROM post-table data table WHERE channel=" a "OR channel=" B "".

Optionally, the query result information includes: at least one query result value.

In some optional implementations of some embodiments, the method further includes:

for the above query result information, the following processing steps are performed:

and a first step of displaying the query result information on a target display terminal in response to determining that the number of the query result values in the at least one query result value satisfies a first selection condition.

The first selection condition is as follows: the number of query result values in the at least one query result value is 1 or less and 0 or more. The target display terminal is a terminal for displaying the query result information in a visual form.

As an example, the above-described query text information may be "average sales of querying a month a items". The query result information may be "500". Therefore, the execution subject can display the query result information on the target display terminal.

And a second step of generating a first visual chart according to the at least one query result value and displaying the first visual chart on the target display terminal in response to determining that the number of the query result values in the at least one query result value meets a second selection condition.

Wherein, the chart types of the first visual chart include: histogram type and line pattern type. Wherein, the second selection condition is: the number of query result values in the at least one query result value is greater than 1 and less than or equal to 10.

By way of example, the above-described query text information may be "average sales of items a every month a of query 2022". The query result information may be "500, 1000, 600, 700, 800, 650, 550, 950, 850, 900". Therefore, the execution subject can display the query result information in the form of a histogram or a line graph on the target display terminal.

And a third step of generating a second visual chart in response to determining that the number of the query result values in the at least one query result value satisfies a third selection condition, and displaying the second visual chart on the target display terminal.

Wherein, the chart type of the second visual chart is a pie chart type. Wherein, the third selection condition is: the number of query result values in the at least one query result value is greater than 10.

By way of example, the query text information may be "average sales of items per day for query 2022, 6, month a". The query result information may be "5, 10, 20, 30, 40, 50, 60, 70, 80, 90,5, 10, 20, 30, 40, 50, 60, 70, 80, 905, 10, 20, 30, 40, 50, 60, 70, 80, 90". Therefore, the inquiry result information is displayed on the target display terminal in the form of a pie chart.

With further reference to FIG. 2, as an implementation of the method shown in the above figures, the present disclosure provides some embodiments of a structured query statement generation device, corresponding to those method embodiments shown in FIG. 1, which may find particular application in a variety of electronic devices.

As shown in fig. 2, the structured query statement generation device 200 of some embodiments includes: an acquisition unit 201, a first generation unit 202, a second generation unit 203, a third generation unit 204, and a data query unit 205. Wherein the obtaining unit 201 is configured to obtain query text information input by a target user and query index definition information corresponding to the query text information, where the query index definition information includes: target index name and index business logic information; the first generating unit 202 is configured to generate candidate index information according to the query text information in response to determining that the target index name is not included in the index name table set in advance; the second generating unit 203 is configured to generate index query information according to a pre-trained index query information generation model, the query text information and the candidate index information, wherein the index query information is structured information; the third generating unit 204 is configured to generate a structured query sentence corresponding to the index query information through a pre-trained structured query sentence generating model and the index business logic information; the data query unit 205 is configured to perform data query on the target database according to the above structured query statement to generate query result information.

It will be appreciated that the elements recited in the structured query statement generation device 200 correspond to the various steps in the method described with reference to figure 1. Thus, the operations, features and advantages described above for the method are equally applicable to the structured query statement generating device 200 and the units contained therein, and are not described here again.

Referring now to fig. 3, a schematic diagram of an electronic device (e.g., computing device) 300 suitable for use in implementing some embodiments of the present disclosure is shown. The electronic device shown in fig. 3 is merely an example and should not impose any limitations on the functionality and scope of use of embodiments of the present disclosure.

As shown in fig. 3, the electronic device 300 may include a processing means (e.g., a central processing unit, a graphics processor, etc.) 301 that may perform various suitable actions and processes in accordance with a program stored in a Read Only Memory (ROM) 302 or a program loaded from a storage means 308 into a Random Access Memory (RAM) 303. In the RAM 303, various programs and data required for the operation of the electronic apparatus 300 are also stored. The processing device 301, the ROM 302, and the RAM 303 are connected to each other via a bus 304. An input/output (I/O) interface 305 is also connected to bus 304.

In general, the following devices may be connected to the I/O interface 305: input devices 306 including, for example, a touch screen, touchpad, keyboard, mouse, camera, microphone, accelerometer, gyroscope, etc.; an output device 307 including, for example, a Liquid Crystal Display (LCD), a speaker, a vibrator, and the like; storage 308 including, for example, magnetic tape, hard disk, etc.; and communication means 309. The communication means 309 may allow the electronic device 300 to communicate with other devices wirelessly or by wire to exchange data. While fig. 3 shows an electronic device 300 having various means, it is to be understood that not all of the illustrated means are required to be implemented or provided. More or fewer devices may be implemented or provided instead. Each block shown in fig. 3 may represent one device or a plurality of devices as needed.

In particular, according to some embodiments of the present disclosure, the processes described above with reference to flowcharts may be implemented as computer software programs. For example, some embodiments of the present disclosure include a computer program product comprising a computer program embodied on a computer readable medium, the computer program comprising program code for performing the method shown in the flow chart. In such embodiments, the computer program may be downloaded and installed from a network via communications device 309, or from storage device 308, or from ROM 302. The above-described functions defined in the methods of some embodiments of the present disclosure are performed when the computer program is executed by the processing means 301.

It should be noted that, the computer readable medium described in some embodiments of the present disclosure may be a computer readable signal medium or a computer readable storage medium, or any combination of the two. The computer readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples of the computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In some embodiments of the present disclosure, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In some embodiments of the present disclosure, however, the computer-readable signal medium may comprise a data signal propagated in baseband or as part of a carrier wave, with the computer-readable program code embodied therein. Such a propagated data signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination of the foregoing. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: electrical wires, fiber optic cables, RF (radio frequency), and the like, or any suitable combination of the foregoing.

In some implementations, the clients, servers may communicate using any currently known or future developed network protocol, such as HTTP (HyperText Transfer Protocol ), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), the internet (e.g., the internet), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future developed networks.

The computer readable medium may be contained in the electronic device; or may exist alone without being incorporated into the electronic device. The computer readable medium carries one or more programs which, when executed by the electronic device, cause the electronic device to: acquiring query text information input by a target user and query index definition information corresponding to the query text information, wherein the query index definition information comprises: target index name and index business logic information; generating candidate index information according to the query text information in response to determining that the target index name is not contained in a preset index name table; generating an index query information according to a pre-trained index query information generation model, the query text information and the candidate index information, wherein the index query information is structured information; generating a structured query statement corresponding to the index query information through a pre-trained structured query statement generation model and the index business logic information; and carrying out data query on the target database according to the structured query statement so as to generate query result information.

Computer program code for carrying out operations for some embodiments of the present disclosure may be written in one or more programming languages, including an object oriented programming language such as Java, smalltalk, C ++ and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computer (for example, through the Internet using an Internet service provider).

The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

The units described in some embodiments of the present disclosure may be implemented by means of software, or may be implemented by means of hardware. The described units may also be provided in a processor, for example, described as: the processor comprises an acquisition unit, a first generation unit, a second generation unit, a third generation unit and a data query unit. The names of these units do not limit the unit itself in some cases, and for example, the acquisition unit may also be described as "a unit that acquires query text information input by the target user and query index definition information corresponding to the query text information" described above.

The functions described above herein may be performed, at least in part, by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: a Field Programmable Gate Array (FPGA), an Application Specific Integrated Circuit (ASIC), an Application Specific Standard Product (ASSP), a system on a chip (SOC), a Complex Programmable Logic Device (CPLD), and the like.

The foregoing description is only of the preferred embodiments of the present disclosure and description of the principles of the technology being employed. It will be appreciated by those skilled in the art that the scope of the invention in the embodiments of the present disclosure is not limited to the specific combination of the above technical features, but encompasses other technical features formed by any combination of the above technical features or their equivalents without departing from the spirit of the invention. Such as the above-described features, are mutually substituted with (but not limited to) the features having similar functions disclosed in the embodiments of the present disclosure.

Claims

1. A structured query statement generation method, comprising:

acquiring query text information input by a target user and query index definition information corresponding to the query text information, wherein the query index definition information comprises: target index name and index business logic information;

generating candidate index information according to the query text information in response to determining that the target index name is not contained in a preset index name table, wherein the index name table comprises: index name collection;

generating a model, the query text information and the candidate index information according to the pre-trained index query information, and generating index query information, wherein the index query information is structured information;

generating a structured query statement corresponding to the index query information through a pre-trained structured query statement generation model and the index business logic information;

performing data query on a target database according to the structured query statement to generate query result information, wherein the generating candidate index information according to the query text information comprises the following steps:

word segmentation processing is carried out on the query text information to obtain a word set;

For each word in the set of words, the following processing steps are performed:

word shielding is carried out on the words included in the query text information, and the query text information after shielding is obtained;

performing text coding on the blocked query text information through a coding model included in a pre-trained part-of-speech determination model to generate coded query text information;

inputting the coded query text information into a decoding model included in the part-of-speech determination model to generate decoded query text information;

part of speech tagging is carried out on the decoded query text information through a pre-trained part of speech tagging model so as to generate part of speech information corresponding to the word;

selecting words with the corresponding part-of-speech information as target part-of-speech information from the word set as candidate words, and obtaining a candidate word set;

for each candidate word in the candidate word set, determining index name similarity of each index name in the candidate word and the index name set as first index name similarity, and obtaining a first index name similarity group;

screening candidate words meeting a first screening condition from the candidate word set to serve as target words, wherein the first screening condition is as follows: the first index name similarity group corresponding to the candidate word has first index name similarity which is larger than the first preset index name similarity and smaller than the second preset index name similarity;

Determining index calculation information corresponding to the target word;

and generating the candidate index information according to the index calculation information and the target word.

2. The method of claim 1, wherein after generating candidate index information from the query text information in response to determining that the target index name is not included in the preset index name table, the method further comprises:

determining index name similarity of the target index name and each index name in the index name set as second index name similarity to obtain a second index name similarity group;

in response to determining that there is a second index name similarity in the second index name similarity group that satisfies a second screening condition, determining the target index name as the candidate index information, wherein the second screening condition is: the second index name similarity is greater than the second preset index name similarity.

3. The method of claim 2, wherein the performing a data query on the target database according to the structured query statement to generate query result information comprises:

analyzing the structured query statement to determine a data table information set to be searched;

Performing table connecting operation on the data table corresponding to the data table information set to be searched in the target database to obtain a data table after table connecting;

and executing the structured query statement, and carrying out data query on the data table after the table connection to generate the query result information.

4. The method of claim 3, wherein the query result information comprises: at least one query result value; and

the method further comprises the steps of:

for the query result information, the following processing steps are executed:

in response to determining that the number of query result values in the at least one query result value meets a first selection condition, displaying the query result information on a target display terminal;

in response to determining that the number of query result values in the at least one query result value meets a second selection condition, generating a first visual chart according to the at least one query result value, and displaying the first visual chart at the target display terminal, wherein a chart type of the first visual chart comprises: histogram type and line pattern type;

and in response to determining that the number of the query result values in the at least one query result value meets a third selection condition, generating a second visual chart, and displaying the second visual chart on the target display terminal, wherein the chart type of the second visual chart is a pie chart type.

5. A structured query statement generation apparatus comprising:

the acquisition unit is configured to acquire query text information input by a target user and query index definition information corresponding to the query text information, wherein the query index definition information comprises: target index name and index business logic information;

a first generation unit configured to generate candidate index information from the query text information in response to determining that the target index name is not included in a preset index name table, wherein the index name table includes: index name collection;

the second generation unit is configured to generate an index query information according to a pre-trained index query information generation model, the query text information and the candidate index information, wherein the index query information is structured information;

the third generation unit is configured to generate a structured query statement corresponding to the index query information through a pre-trained structured query statement generation model and the index business logic information;

the data query unit is configured to perform data query on a target database according to the structured query statement to generate query result information, wherein the generating candidate index information according to the query text information comprises the following steps:

Determining index calculation information corresponding to the target word;

6. An electronic device, comprising:

one or more processors;

a storage device having one or more programs stored thereon;

when executed by the one or more processors, causes the one or more processors to implement the method of any of claims 1 to 4.

7. A computer readable medium having stored thereon a computer program, wherein the program when executed by a processor implements the method of any of claims 1 to 4.