Intelligent review method and system

Abstract

The application discloses an intelligent evaluation method and system, and relates to the technical field of data processing.The intelligent evaluation system comprises a cloud server, a terminal server and a user terminal; a database containing a plurality of sample bid texts is acquired, sample subtexts are extracted from the sample bid texts according to a plurality of preset evaluation contents; an evaluation module matched with the evaluation content represented by the sample subtexts is selected from an evaluation model, and the sample subtexts are evaluated and processed by using the evaluation module to obtain sample subevaluation values; sample features are obtained by performing feature extraction on the sample subtexts, sample subtexts representing the same evaluation content are processed in groups to obtain a plurality of sample subtext groups, two sample subtexts representing the same evaluation content are selected, and the evaluation model is adjusted in parameters according to the relationship between the sample subevaluation values of the two sample subtexts and is then sent to the terminal server, so that the intelligence and stability of the evaluation quality can be improved.

Description

An intelligent review method and system Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to an intelligent review method and system. Background Technology

[0002] Tendering and bidding is a comprehensive economic responsibility system that promotes competition in the basic construction sector. It is a transaction method used in a market economy for the buying and selling of bulk goods, the contracting and subcontracting of engineering construction projects, and the procurement and provision of services.

[0003] Bid evaluation (i.e., reviewing bid documents) is the core of the bidding process. In this stage, the bid evaluation committee reviews and compares the bid documents according to the evaluation criteria and methods specified in the bidding documents to determine the most suitable bidder. The evaluation process is not only the foundation for the successful implementation of the project but also a guarantee for improving the resource utilization efficiency of the bidding entity.

[0004] Intelligent review systems utilize artificial intelligence technology to simulate the thinking and standards of professional reviewers, enabling rapid and objective evaluation of various content requiring review. However, intelligent review systems often suffer from inconsistent review quality. Therefore, effectively improving the stability of the review quality of intelligent review systems has become a pressing issue. Summary of the Invention

[0005] This specification provides an intelligent review method and system to offer a review scheme that effectively improves the stability of review quality.

[0006] In a first aspect, embodiments of this specification provide an intelligent review system, which includes a cloud server, a terminal server, and a user terminal; the cloud server is used for:

[0007] Upon receiving a model deployment instruction, a database containing multiple sample bid texts is obtained, and sample sub-texts are extracted from the sample bid texts according to multiple preset review contents.

[0008] From the review model, select a review module that matches the review content represented by the sample sub-text, and use the review module to review the sample sub-text to obtain the sample sub-review value;

[0009] Feature extraction is performed on the sample subtexts to obtain sample features, and based on the sample features, sample subtexts representing the same review content are grouped to obtain multiple sample subtext groups.

[0010] Based on the sample subtext group, two sample subtexts representing the same review content are selected, and the review model is adjusted according to the relationship between the sample sub-review values ​​of the two sample subtexts before being sent to the terminal server.

[0011] The user terminal is used to send the tender text to be reviewed to the terminal server;

[0012] The terminal server is used to review the bid text sent by the user terminal after receiving and deploying the review model sent by the cloud server, obtain the review result, and send the review result to the user terminal.

[0013] Secondly, embodiments of this specification provide an intelligent review method for a cloud server, comprising:

[0014] Upon receiving a model deployment instruction, a database containing multiple sample bid texts is obtained, and sample sub-texts are extracted from the sample bid texts according to multiple preset review contents.

[0015] From the review model, select a review module that matches the review content represented by the sample sub-text, and use the review module to review the sample sub-text to obtain the sample sub-review value;

[0016] Feature extraction is performed on the sample subtexts to obtain sample features, and based on the sample features, sample subtexts representing the same review content are grouped to obtain multiple sample subtext groups.

[0017] Based on the sample subtext group, two sample subtexts representing the same review content are selected. According to the relationship between the sample sub-review values ​​of the two sample subtexts, the review model is adjusted by parameters and then sent to the terminal server. After receiving and deploying the review model sent by the cloud server, the terminal server reviews the bid text sent by the user terminal, obtains the review result, and sends the review result to the user terminal.

[0018] Thirdly, embodiments of this specification provide an intelligent review method for a terminal server, comprising:

[0019] After receiving and deploying the review model sent by the cloud server, the bidding text sent by the user terminal is reviewed to obtain the review result. The review model is obtained by the cloud server after adjusting the parameters of the review model based on the relationship between the sample sub-review values ​​of two sample sub-texts. The two sample sub-texts are obtained by the cloud server selecting sample sub-texts representing the same review content based on the sample sub-text group. The sample sub-text group is obtained by the cloud server grouping sample sub-texts representing the same review content based on sample features. The sample features are obtained by the cloud server extracting features from the sample sub-texts. The sample sub-review value is obtained by the cloud server using the review module to review the sample sub-text. The review module is the module selected by the cloud server from the review model that matches the review content represented by the sample sub-text. The sample sub-texts are extracted by the cloud server from the sample bidding texts based on multiple preset review contents. The sample bidding texts are the texts contained in the database obtained by the cloud server when it receives the model deployment instruction.

[0020] Fourthly, embodiments of this specification provide an electronic device comprising: a processor, and a memory arranged to store computer-executable instructions, wherein, when the executable instructions are executed, the processor is capable of:

[0021] Upon receiving a model deployment instruction, a database containing multiple sample bid texts is obtained, and sample sub-texts are extracted from the sample bid texts according to multiple preset review contents.

[0022] From the review model, select a review module that matches the review content represented by the sample sub-text, and use the review module to review the sample sub-text to obtain the sample sub-review value;

[0023] Feature extraction is performed on the sample subtexts to obtain sample features, and based on the sample features, sample subtexts representing the same review content are grouped to obtain multiple sample subtext groups.

[0024] Based on the sample subtext group, two sample subtexts representing the same review content are selected. According to the relationship between the sample sub-review values ​​of the two sample subtexts, the review model is adjusted by parameters and then sent to the terminal server. After receiving and deploying the review model sent by the cloud server, the terminal server reviews the bid text sent by the user terminal, obtains the review result, and sends the review result to the user terminal.

[0025] Fifthly, embodiments of this specification provide an electronic device comprising: a processor, and a memory arranged to store computer-executable instructions, wherein, when the executable instructions are executed, the processor is capable of:

[0026] After receiving and deploying the review model sent by the cloud server, the bidding text sent by the user terminal is reviewed to obtain the review result. The review model is obtained by the cloud server after adjusting the parameters of the review model based on the relationship between the sample sub-review values ​​of two sample sub-texts. The two sample sub-texts are obtained by the cloud server selecting sample sub-texts representing the same review content based on the sample sub-text group. The sample sub-text group is obtained by the cloud server grouping sample sub-texts representing the same review content based on sample features. The sample features are obtained by the cloud server extracting features from the sample sub-texts. The sample sub-review value is obtained by the cloud server using the review module to review the sample sub-text. The review module is the module selected by the cloud server from the review model that matches the review content represented by the sample sub-text. The sample sub-texts are extracted by the cloud server from the sample bidding texts based on multiple preset review contents. The sample bidding texts are the texts contained in the database obtained by the cloud server when it receives the model deployment instruction.

[0027] The review results are sent to the user terminal.

[0028] Sixthly, embodiments of this specification provide a storage medium for storing a computer program, the computer program being executable by a processor to implement the following processes:

[0029] Upon receiving a model deployment instruction, a database containing multiple sample bid texts is obtained, and sample sub-texts are extracted from the sample bid texts according to multiple preset review contents.

[0030] From the review model, select a review module that matches the review content represented by the sample sub-text, and use the review module to review the sample sub-text to obtain the sample sub-review value;

[0031] Feature extraction is performed on the sample subtexts to obtain sample features, and based on the sample features, sample subtexts representing the same review content are grouped to obtain multiple sample subtext groups.

[0032] Based on the sample subtext group, two sample subtexts representing the same review content are selected. According to the relationship between the sample sub-review values ​​of the two sample subtexts, the review model is adjusted by parameters and then sent to the terminal server. After receiving and deploying the review model sent by the cloud server, the terminal server reviews the bid text sent by the user terminal, obtains the review result, and sends the review result to the user terminal.

[0033] Seventhly, embodiments of this specification provide a storage medium for storing a computer program, the computer program being executable by a processor to implement the following processes:

[0034] After receiving and deploying the review model sent by the cloud server, the bidding text sent by the user terminal is reviewed to obtain the review result. The review model is obtained by the cloud server after adjusting the parameters of the review model based on the relationship between the sample sub-review values ​​of two sample sub-texts. The two sample sub-texts are obtained by the cloud server selecting sample sub-texts representing the same review content based on the sample sub-text group. The sample sub-text group is obtained by the cloud server grouping sample sub-texts representing the same review content based on sample features. The sample features are obtained by the cloud server extracting features from the sample sub-texts. The sample sub-review value is obtained by the cloud server using the review module to review the sample sub-text. The review module is the module selected by the cloud server from the review model that matches the review content represented by the sample sub-text. The sample sub-texts are extracted by the cloud server from the sample bidding texts based on multiple preset review contents. The sample bidding texts are the texts contained in the database obtained by the cloud server when it receives the model deployment instruction.

[0035] The review results are sent to the user terminal.

[0036] Beneficial Effects: Compared to existing technologies, this invention, after extracting sample sub-texts from sample bid texts, first uses a review module to obtain sample sub-review values. Then, it groups the sample sub-texts according to their characteristics, creating sample sub-text groups. Next, it selects two sample sub-texts from each group. Based on the relationship between the sample sub-review values ​​of these two sub-texts, the parameters of the review model can be adjusted, and the review model can be deployed on a terminal server. This process, by adjusting the parameters of the review model through the relationship between the sample sub-review values ​​of two sample sub-texts representing the same review content, improves the accuracy of the review results obtained by the review model deployed on the terminal server, thereby enhancing the stability of the review quality of the intelligent review system in which the review model resides.

[0037] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this specification.

[0038] Other features and aspects of this specification will become clear from the following detailed description of exemplary embodiments with reference to the accompanying drawings. Attached Figure Description

[0039] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with this specification and, together with the specification, serve to explain the technical solutions described herein.

[0040] Figure 1 shows a block diagram of an evaluation system according to an embodiment of this specification.

[0041] Figure 2 shows a flowchart of an intelligent review method according to an embodiment of this specification.

[0042] Figure 3 shows a flowchart of an intelligent review method for a cloud server according to an embodiment of this specification.

[0043] Figure 4 shows a flowchart of an intelligent review method for a terminal server according to another embodiment of this specification.

[0044] Figure 5 shows a block diagram of an electronic device according to an embodiment of this specification.

[0045] Figure 6 shows a block diagram of another electronic device according to an embodiment of this specification. Detailed Implementation

[0046] Various exemplary embodiments, features, and aspects of this specification will now be described in detail with reference to the accompanying drawings. The same reference numerals in the drawings denote elements that have the same or similar functions. Although various aspects of the embodiments are shown in the drawings, they are not necessarily drawn to scale unless specifically indicated otherwise.

[0047] The term “exemplary” as used herein means “serving as an example, embodiment, or illustration.” Any embodiment illustrated herein as “exemplary” is not necessarily to be construed as superior to or better than other embodiments.

[0048] In this document, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent three cases: A alone, A and B simultaneously, and B alone. Furthermore, the term "at least one" in this document means any combination of at least two of any one or more elements. For example, including at least one of A, B, and C can mean including any one or more elements selected from the set consisting of A, B, and C.

[0049] Furthermore, to better illustrate this specification, numerous specific details are provided in the following detailed embodiments. Those skilled in the art will understand that this specification can be practiced without certain specific details. In some instances, methods, means, components, and circuits well known to those skilled in the art have not been described in detail in order to highlight the main points of this specification.

[0050] Figure 1 illustrates a block diagram of a review system; as shown in Figure 1, the review system includes: a cloud server, a terminal server, and user terminals. Figure 2 shows a flowchart of the intelligent review method corresponding to this review system; as shown in Figure 2, the intelligent review method includes:

[0051] In step S201, upon receiving a model deployment instruction, the cloud server obtains a database containing multiple sample bid texts and extracts sample sub-texts from the sample bid texts according to multiple preset review criteria.

[0052] The model deployment instruction can be a command sent by a server (such as a cloud server or other server) that manages the deployment and / or updating of the review model. The model deployment instruction is a command for generating and / or updating the review model, which is a model used to review tender documents. The model deployment instruction can be a periodically sent instruction, meaning that after the review model is generated by the cloud server and deployed to the terminal server, the cloud server can periodically update the review model.

[0053] The database contains multiple sample bid documents, which can be historical bid documents. Specifically, the database can contain various types of bid documents, such as construction bid documents and design bid documents, allowing the resulting evaluation model to be used in multiple fields. Alternatively, the database can contain only one type of bid document, such as only building construction bid documents, allowing the resulting evaluation model to be used exclusively in the building construction field. This specification does not impose specific limitations on the number and type of sample bid documents in the database; these can be determined based on actual circumstances.

[0054] The review content consists of specific sections of the sample bid document, and the sum of all review content sections constitutes the sample bid document. Specifically, the review content can be determined based on the bid document's domain or application scenario. In one example, the bid document can be divided into multiple review sections such as a scheme overview, construction schedule, and management personnel configuration, based on its title. Sample sub-texts are the text corresponding to the review content within the sample bid document. After extracting the review content from the sample bid document, each sample sub-text corresponding to each review section can be further reviewed separately. This modular and targeted processing of the sample bid document simplifies the complex issue of bid document review while ensuring accurate review of each sample sub-text, thus improving the accuracy of the review results.

[0055] In step S202, the cloud server selects a review module from the review model that matches the review content represented by the sample sub-text, and uses the review module to review the sample sub-text to obtain the sample sub-review value.

[0056] The review module is a module within the review model. The review model can contain multiple review modules, each capable of reviewing at least one sub-text (a portion of the text in the bid document) corresponding to a review item. The sample sub-text is a portion of the sample bid document.

[0057] A single review item can have multiple review metrics. For example, for a review item like "solution overview," the review metrics could include logical metrics (such as clear and coherent logic), textual consistency (such as consistent descriptions), and thematic metrics (such as high completeness). Each review metric uses a different natural language processing technique; for instance, the thematic metric might use entity recognition technology, while textual consistency might use text consistency detection technology. Furthermore, different review items may have different review metrics, making the generation process of the review model quite complex. It should be understood that setting the review model generation process on a powerful cloud server can improve the accuracy of the generated review model, while setting the review model usage process on a high-real-time terminal server can improve the real-time performance of the review process.

[0058] As mentioned earlier, different review content may correspond to different review metrics. If the same review strategy is used for different review content, the effect may be compromised. To improve the review effect, a review module can be set up for each review content. In this way, the number of review modules in the review model can be the same as the number of review content. In one example, the corresponding natural language processing technology can be determined based on the various review metrics corresponding to the review content, and then the review module can be constructed using this processing technology. The construction process of the review module will not be described in detail in this specification.

[0059] When a sample bid text is divided into multiple sample subtexts, a corresponding review module can be determined for each sample subtext. That is, the review module can be determined based on the review content represented by the sample subtext. After the review module is determined, it can be used to review the corresponding sample subtext and obtain the review score (i.e., the sample sub-review value) of the sample subtext.

[0060] In step S203, the cloud server extracts features from the sample sub-texts to obtain sample features, and based on the sample features, groups the sample sub-texts representing the same review content to obtain multiple sample sub-text groups.

[0061] In step S204, the cloud server selects two sample sub-texts representing the same review content based on the sample sub-text group, and adjusts the parameters of the review model according to the relationship between the sample sub-review values ​​of the two sample sub-texts before sending it to the terminal server.

[0062] Two sample subtexts representing the same review content may not necessarily convey the same information. For example, regarding the review content on management personnel allocation, the personnel information in one sample subtext may be consistent, while the personnel information in another sample subtext may be problematic. When the information conveyed by the texts differs, the sample sub-review values ​​of the two sample subtexts should be different; conversely, when the information conveyed by the texts is the same, the sample sub-review values ​​of the two sample subtexts should be similar.

[0063] In one example, the review model can be trained based on the relationship between the sample sub-review values ​​corresponding to two sample sub-texts representing the same review content. Since whether two sample sub-texts representing the same review content express the same information can be determined by whether the features extracted from them are consistent, the review model can be trained based on the relationship between the sample sub-review values ​​corresponding to sample sub-texts with the same sample features, or the relationship between the sample sub-review values ​​corresponding to sample sub-texts with different sample features.

[0064] Specifically, features can be extracted from each sample sub-text to obtain sample features. Based on these features, sample sub-texts representing the same review content can be grouped to obtain multiple sample sub-text groups. After grouping, sample sub-texts within the same group have the same expressive information, while sample sub-texts outside the same group have different expressive information. This is reflected in the sample sub-review values: sample sub-review values ​​within the same group are similar, while sample sub-review values ​​outside the same group are not similar.

[0065] Therefore, two sample subtexts representing the same review content can be selected from the sample subtext group. These two sample subtexts may be in the same sample subtext group or in two different sample subtext groups. The parameters of the review model can be adjusted based on the relationship between the sample sub-review values ​​of the two sample subtexts. Specifically, the parameters of the review module corresponding to the sample subtext in the review model can be adjusted based on the relationship between the sample sub-review values ​​of the two sample subtexts.

[0066] In one example, the relationship is a difference or ratio. When selecting two sample sub-texts representing the same review content based on the sample sub-text group, and adjusting the parameters of the review model according to the relationship between the sample sub-review values ​​of the two sample sub-texts, the cloud server is used for:

[0067] From the same sample subtext group, select two sample subtexts and obtain the review module that matches the review content represented by the sample subtexts;

[0068] If the difference or ratio between the sample sub-review values ​​of two sample sub-texts exceeds a first preset value, the parameters of the review module corresponding to the sample sub-text are adjusted.

[0069] Specifically, since the review content represented by the sample sub-texts in the same sample sub-text group is the same and the sample characteristics are similar, the difference or ratio of the sample sub-review values ​​corresponding to any two sample sub-texts in the same sample sub-text group should be similar. Therefore, if the difference or ratio of any two sample sub-texts in the same sample sub-text group exceeds a first preset value, the parameters of the review module corresponding to the review content represented by the sample sub-texts in the sample sub-text group can be adjusted so that the review module after parameter adjustment outputs similar sample sub-review values ​​for the sample sub-texts in the same sample sub-text group. This can improve the accuracy and stability of the review model.

[0070] The relationships between sample sub-evaluation values ​​can be expressed not only by magnitude (such as the relationship between differences or ratios), but also by more complex formulas. This specification does not impose specific limitations on the relationships between sample sub-evaluation values; these relationships can be determined based on the actual situation.

[0071] In step S205, the user terminal sends the tender text to be reviewed to the terminal server.

[0072] In step S206, after receiving and deploying the review model sent by the cloud server, the terminal server reviews the bid text sent by the user terminal, obtains the review result, and sends the review result to the user terminal.

[0073] The terminal server is the server that uses the review model. This terminal server can connect to user terminals and receive the tender documents to be reviewed from the user terminals. After the review parameters of the review model are adjusted, the review model can be deployed to the terminal server, which can then use the review model to review the tender documents sent by the user terminals.

[0074] In the embodiments of this specification, after the cloud server extracts sample sub-texts from the sample bid text, it first uses the review module to obtain sample sub-review values. Then, it groups the sample sub-texts according to their sample characteristics to obtain sample sub-text groups. Next, it selects two sample sub-texts from each sample sub-text group. Based on the relationship between the sample sub-review values ​​of these two sample sub-texts, it adjusts the parameters of the review model and deploys the review model on the terminal server. This process, by adjusting the parameters of the review model through the relationship between the sample sub-review values ​​of two sample sub-texts representing the same review content, can improve the accuracy of the review results obtained by the review model deployed on the terminal server, thereby improving the stability of the review quality of the intelligent review system in which the review model resides.

[0075] The preceding example illustrates how to adjust the parameters of a review model using sample sub-review values ​​from sample subtexts within the same sample subtext group. Similarly, sample sub-review values ​​from sample subtexts from different sample subtext groups can also be used to adjust the parameters of the review model. In one implementation, the relationship is a difference or ratio. When selecting two sample subtexts representing the same review content based on the sample subtext group, and adjusting the parameters of the review model according to the relationship between the sample sub-review values ​​of the two sample subtexts, the cloud server is used to:

[0076] Obtain the first descriptive text of the sample sub-text group;

[0077] From multiple sample subtext groups representing the same review content, two target sample subtext groups are obtained, and there are opposite descriptions in the first description text of the two target sample subtext groups.

[0078] Select one sample subtext from each of the two target sample subtext groups, and adjust the parameters of the review module matched with the target sample subtext group if the difference or ratio between the sample sub-review values ​​of the two sample subtexts is less than a second preset value.

[0079] The first descriptive text is used to describe the common features of the sample subtexts in the sample subtext group.

[0080] As mentioned earlier, the same review content can correspond to multiple review indicators. In one example, the text of a sample sub-text group under each review indicator can be generated based on the review indicators, and the texts under each review indicator can be integrated to obtain the first descriptive text. In one implementation, when obtaining the first descriptive text of the sample sub-text group, the cloud server is used to:

[0081] The sample sub-text groups are used as text data for a pre-defined large language model.

[0082] The review indicators contained in the review content represented by the sample subtext are used as format guidance data for inputting into a large language model. The review indicators are indicators that specify the review perspective under the review content.

[0083] The text data and the format guidance data are input into the large language model, and the text output by the large language model based on the text data and the format guidance data is used as the first descriptive text.

[0084] Large Language Models (LMMs) refer to neural network models with billions or even trillions of parameters, which have powerful functions such as text understanding, generation, and reasoning.

[0085] In one example, a large language model can be used in the generation of the first descriptive text. Specifically, all or part of the sample subtexts from the sample subtext group can be input into the large language model as input text data; the review metrics contained in the review content described by the sample subtexts can also be input into the large language model as guiding data for the output of the large language model. This ensures that the text output by the large language model contains data corresponding to each review metric. This example, by using a large language model, yields a relatively accurate first descriptive text.

[0086] After obtaining the first description text, the first description texts of each sample subtext group representing the same review content can be compared to obtain two first description texts with opposite description content. The sample subtext group corresponding to these two first description texts is the target sample subtext group.

[0087] Since the sample subtexts from different target sample subtext groups represent the same review content but have different sample features, the difference or ratio of the sample sub-review values ​​corresponding to the sample subtexts from different target sample subtext groups should be relatively large. Therefore, if the difference or ratio of the sample sub-review values ​​corresponding to the sample subtexts from different target sample subtext groups is less than a second preset value, the parameters of the review module corresponding to the target sample subtext group can be adjusted so that the adjusted review module outputs sample sub-review values ​​with a larger difference for the sample subtexts from different target sample subtext groups. This can improve the accuracy and stability of the review model.

[0088] As mentioned earlier, the same review content may contain multiple review indicators. Therefore, it may be impossible to find a target sample subtext group in which the texts corresponding to each review indicator are all opposite in the first description text. In one implementation, when obtaining two target sample subtext groups from multiple sample subtext groups representing the same review content, the cloud server is used to:

[0089] Obtain the weight value of each review indicator in the review content, and take the review indicator corresponding to the weight value that is greater than the third preset value as the target review indicator.

[0090] From multiple sample sub-text groups representing the same review content, select two sample sub-text groups whose descriptions of the target review indicators in the first description text are opposite to each other, and use them as the target sample sub-text groups.

[0091] The weight value is used to indicate the importance of the review indicator in the review content.

[0092] Specifically, the weight values ​​of each review indicator in the same review content can be obtained first, and a third preset value related to the weight value can be preset. In this way, the review indicators corresponding to the weight values ​​that are greater than the third preset value can be used as the review indicators (i.e. target review indicators) required to determine the target sample sub-text group in the review content.

[0093] The importance of the weight values ​​can be determined by relevant experts based on their experience, or it can be determined based on the descriptive text of the review indicators. In one implementation, when obtaining the weight values ​​of each review indicator in the review content, the cloud server is used for:

[0094] Obtain the key words corresponding to the review content and the importance scores of the key words;

[0095] Extract the key words from the second descriptive text of the evaluation indicators;

[0096] The importance score of the review indicator is determined based on the frequency and importance score of the marker words appearing in the second description text.

[0097] The importance scores of each evaluation indicator in the evaluation content are normalized to obtain the weight value of each evaluation indicator in the evaluation content.

[0098] In this context, "signature words" refers to the key terms representing the content being reviewed, and "importance score" is the score obtained by measuring the importance of these signature words. The signature words and importance scores can be determined by relevant experts based on their experience. Different review content should correspond to different signature words and / or importance scores.

[0099] The second descriptive text is the text that describes the review indicators. After obtaining the key words and importance scores, key words can be extracted from each second descriptive text. Based on the frequency of occurrence and importance scores of the key words in the second descriptive text, the importance score of the review indicator corresponding to that second descriptive text is determined. For example, in the second descriptive text of a review indicator, the key words are A, B, and C. The importance score of key word A is a, the importance score of key word B is b, and the importance score of key word C is c. Then, the importance score of the review indicator F = a*A + b*B + c*C.

[0100] After obtaining the importance scores of each review indicator in the review content, these scores can be normalized to obtain the weight value of each review indicator.

[0101] It should be understood that, to avoid the method failing, the second descriptive text should be concise and highlight the key points. Specifically, large-scale language models and other methods can be used to obtain the second descriptive text.

[0102] The above process provides a method for determining the weight value of the review indicator based on the marker words and importance scores. Since the frequency of occurrence of the marker words and the importance scores of the marker words can reflect the importance of the review indicator, the obtained weight value has high accuracy.

[0103] The target review metric is the review metric used to determine the target sample sub-text group. Specifically, after obtaining the weight values, only the review metrics corresponding to weight values ​​greater than a third preset value can be used as the target review metrics. In this way, among multiple sample sub-text groups representing the same review content, two target sample sub-text groups can be selected based on the description content corresponding to the target review metric. In the first description text corresponding to sample sub-texts from different target sample sub-text groups, the description content corresponding to the target review metric is reversed.

[0104] In the embodiments of this specification, target review indicators are determined based on weight values, thereby enabling the determination of target sample sub-text groups using only the target review values. Since the target review indicators are review indicators with relatively large weight values ​​in the review content, this process can improve the speed of determining target sample sub-text groups while maintaining the accuracy of the determination.

[0105] In one implementation, when reviewing the tender text sent by the user terminal to obtain a review result, the terminal server is configured to:

[0106] Upon receiving the bid text sent by the user terminal, subtext is extracted from the bid text based on multiple preset review criteria.

[0107] From the review models deployed on the cloud server, select a review module that matches the review content represented by the sub-text, and use the review module to review the sub-text to obtain a sub-review value;

[0108] The sub-evaluation values ​​corresponding to each sub-text in the bid document are summed to obtain the evaluation value of the bid document, and the evaluation result is obtained based on the evaluation value.

[0109] Similar to the training process of the aforementioned review model, when using the review model, the terminal server can first segment the bid text based on the preset review content to obtain multiple sub-texts, and then use the review module in the review model that matches the review content to review the sub-texts and obtain sub-review values.

[0110] Furthermore, the sub-review values ​​can be summed to obtain the review value sent to the user terminal. To further differentiate the importance of each review item, weights can be assigned to each sub-review value, and the sub-review values ​​can be summed according to the weights. In addition to the review value, the review results can also include sub-review values ​​and other content, making it easier for users of the user terminal to understand the details of the tender document.

[0111] In the embodiments of this specification, after extracting the bid text through the review content, the sub-texts corresponding to each review content can be subjected to subsequent review processing, realizing modular and targeted processing of the bid text. While simplifying the complex problem of bid text review, it can also ensure that each sub-text can be accurately reviewed, which helps to improve the accuracy of the review results.

[0112] Figure 3 illustrates a flowchart of an intelligent review method for use on a cloud server; as shown in Figure 3, the intelligent review method includes:

[0113] Step S301: Upon receiving a model deployment instruction, obtain a database containing multiple sample bid texts, and extract sample sub-texts from the sample bid texts according to multiple preset review contents.

[0114] Step S302: Select a review module from the review model that matches the review content represented by the sample sub-text, and use the review module to review the sample sub-text to obtain the sample sub-review value.

[0115] Step S303: Extract features from the sample sub-texts to obtain sample features, and based on the sample features, group sample sub-texts representing the same review content to obtain multiple sample sub-text groups.

[0116] Step S304: Based on the sample sub-text group, select two sample sub-texts representing the same review content, and adjust the parameters of the review model according to the relationship between the sample sub-review values ​​of the two sample sub-texts before sending it to the terminal server. This allows the terminal server to review the bid text sent by the user terminal after receiving and deploying the review model sent by the cloud server, obtain the review result, and send the review result to the user terminal.

[0117] In the embodiments of this specification, after the cloud server extracts sample sub-texts from the sample bid text, it first uses the review module to obtain sample sub-review values. Then, it groups the sample sub-texts according to their sample characteristics to obtain sample sub-text groups. Next, it selects two sample sub-texts from each sample sub-text group. Based on the relationship between the sample sub-review values ​​of these two sample sub-texts, it adjusts the parameters of the review model and deploys the review model on the terminal server. This process, by adjusting the parameters of the review model through the relationship between the sample sub-review values ​​of two sample sub-texts representing the same review content, can improve the accuracy of the review results obtained by the review model deployed on the terminal server, thereby improving the stability of the review quality of the intelligent review system in which the review model resides.

[0118] Figure 4 illustrates a flowchart of an intelligent review method for a terminal server; as shown in Figure 4, the intelligent review method includes:

[0119] Step S401: After receiving and deploying the review model sent by the cloud server, the bidding text sent by the user terminal is reviewed to obtain the review result. The review model is obtained by the cloud server after adjusting the parameters of the review model based on the relationship between the sample sub-review values ​​of two sample sub-texts. The two sample sub-texts are obtained by the cloud server selecting sample sub-texts representing the same review content based on the sample sub-text group. The sample sub-text group is obtained by the cloud server grouping sample sub-texts representing the same review content based on sample features. The sample features are obtained by the cloud server extracting features from the sample sub-texts. The sample sub-review value is obtained by the cloud server using the review module to review the sample sub-text. The review module is the module selected by the cloud server from the review model that matches the review content represented by the sample sub-text. The sample sub-texts are extracted by the cloud server from the sample bidding texts based on multiple preset review contents. The sample bidding texts are the texts contained in the database obtained by the cloud server when it receives the model deployment instruction.

[0120] Step S401: Send the review result to the user terminal.

[0121] In the embodiments of this specification, after the cloud server extracts sample sub-texts from the sample bid text, it first uses the review module to obtain sample sub-review values. Then, it groups the sample sub-texts according to their sample characteristics to obtain sample sub-text groups. Next, it selects two sample sub-texts from each sample sub-text group. Based on the relationship between the sample sub-review values ​​of these two sample sub-texts, it adjusts the parameters of the review model and deploys the review model on the terminal server. This process, by adjusting the parameters of the review model through the relationship between the sample sub-review values ​​of two sample sub-texts representing the same review content, can improve the accuracy of the review results obtained by the review model deployed on the terminal server, thereby improving the stability of the review quality of the intelligent review system in which the review model resides.

[0122] It is understood that the various method embodiments mentioned above in this specification can be combined with each other to form combined embodiments without violating the underlying principles and logic. Due to space limitations, these will not be elaborated upon further in this specification. Those skilled in the art will understand that the specific execution order of each step in the above methods of specific implementation should be determined by its function and possible internal logic.

[0123] In addition, this specification also provides electronic devices, computer-readable storage media, and programs, all of which can be used to implement any of the intelligent review methods provided in this specification. The corresponding technical solutions and descriptions are described in the relevant sections of the Methods section and will not be repeated here.

[0124] This specification also provides an embodiment of a computer-readable storage medium storing computer program instructions that, when executed by a processor, implement the above-described method. The computer-readable storage medium may be a non-volatile computer-readable storage medium.

[0125] This specification also provides an electronic device comprising: a processor; and a memory for storing processor-executable instructions; wherein the processor is configured to invoke the instructions stored in the memory to execute the above-described method.

[0126] This specification also provides a computer program product including computer-readable code, which, when run on a device, causes a processor in the device to execute instructions for implementing the image segmentation method provided in any of the above embodiments.

[0127] This specification also provides another computer program product for storing computer-readable instructions that, when executed, cause a computer to perform the operation of the image segmentation method provided in any of the above embodiments.

[0128] Electronic devices can be provided as terminals, servers, or other forms of devices.

[0129] Figure 5 shows a block diagram of an electronic device 800 according to an embodiment of this specification. For example, the electronic device 800 may be a mobile phone, computer, digital broadcasting terminal, messaging device, game console, tablet device, medical device, fitness equipment, personal digital assistant, and other terminals.

[0130] Referring to FIG5, the electronic device 800 may include one or more of the following components: processing component 802, memory 804, power supply component 806, multimedia component 808, audio component 810, input / output (I / O) interface 812, sensor component 814, and communication component 816.

[0131] Processing component 802 typically controls the overall operation of electronic device 800, such as operations associated with display, telephone calls, data communication, camera operation, and recording operations. Processing component 802 may include one or more processors 820 to execute instructions to complete all or part of the steps of the methods described above. Furthermore, processing component 802 may include one or more modules to facilitate interaction between processing component 802 and other components. For example, processing component 802 may include a multimedia module to facilitate interaction between multimedia component 808 and processing component 802.

[0132] Memory 804 is configured to store various types of data to support the operation of electronic device 800. Examples of such data include instructions for any application or method operating on electronic device 800, contact data, phonebook data, messages, pictures, videos, etc. Memory 804 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0133] Power supply component 806 provides power to various components of electronic device 800. Power supply component 806 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to electronic device 800.

[0134] Multimedia component 808 includes a screen that provides an output interface between the electronic device 800 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touchscreen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensors may sense not only the boundaries of the touch or swipe action but also the duration and pressure associated with the touch or swipe operation. In some embodiments, multimedia component 808 includes a front-facing camera and / or a rear-facing camera. When the electronic device 800 is in an operating mode, such as a shooting mode or a video mode, the front-facing camera and / or the rear-facing camera may receive external multimedia data. Each front-facing camera and rear-facing camera may be a fixed optical lens system or have focal length and optical zoom capabilities.

[0135] Audio component 810 is configured to output and / or input audio signals. For example, audio component 810 includes a microphone (MIC) configured to receive external audio signals when electronic device 800 is in an operating mode, such as call mode, recording mode, and voice recognition mode. The received audio signals may be further stored in memory 804 or transmitted via communication component 816. In some embodiments, audio component 810 also includes a speaker for outputting audio signals.

[0136] I / O interface 812 provides an interface between processing component 802 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to, home buttons, volume buttons, power buttons, and lock buttons.

[0137] Sensor assembly 814 includes one or more sensors for providing state assessments of various aspects of electronic device 800. For example, sensor assembly 814 can detect the on / off state of electronic device 800, the relative positioning of components such as the display and keypad of electronic device 800, changes in position of electronic device 800 or a component of electronic device 800, the presence or absence of user contact with electronic device 800, orientation or acceleration / deceleration of electronic device 800, and temperature changes of electronic device 800. Sensor assembly 814 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. Sensor assembly 814 may also include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, sensor assembly 814 may also include an accelerometer, gyroscope, magnetometer, pressure sensor, or temperature sensor.

[0138] Communication component 816 is configured to facilitate wired or wireless communication between electronic device 800 and other devices. Electronic device 800 can access wireless networks based on communication standards, such as WiFi, 2G, or 3G, or combinations thereof. In one exemplary embodiment, communication component 816 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, communication component 816 also includes a near-field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0139] In an exemplary embodiment, the electronic device 800 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the methods described above.

[0140] In an exemplary embodiment, a non-volatile computer-readable storage medium is also provided, such as a memory 804 including computer program instructions that can be executed by a processor 820 of an electronic device 800 to perform the above-described method.

[0141] Figure 6 shows a block diagram of an electronic device 1900 according to an embodiment of this specification. For example, the electronic device 1900 may be provided as a server. Referring to Figure 6, the electronic device 1900 includes a processing component 1922, which further includes one or more processors, and memory resources represented by memory 1932 for storing instructions executable by the processing component 1922, such as application programs. The application programs stored in memory 1932 may include one or more modules, each corresponding to a set of instructions. Furthermore, the processing component 1922 is configured to execute instructions to perform the methods described above.

[0142] Electronic device 1900 may also include a power supply component 1926 configured to perform power management of electronic device 1900, a wired or wireless network interface 1950 configured to connect electronic device 1900 to a network, and an input / output (I / O) interface 1958. Electronic device 1900 can operate on an operating system, such as Windows Server, stored in memory 1932. TM Mac OS X TM Unix™, Linux TM FreeBSD TM Or similar.

[0143] In an exemplary embodiment, a non-volatile computer-readable storage medium is also provided, such as a memory 1932 including computer program instructions that can be executed by a processing component 1922 of an electronic device 1900 to perform the above-described method.

[0144] This specification may be a system, method, and / or computer program product. A computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for causing a processor to implement various aspects of this specification.

[0145] Computer-readable storage media can be tangible devices capable of holding and storing instructions for use by an instruction execution device. Computer-readable storage media can be, for example—but not limited to—electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or recessed protrusions storing instructions thereon, and any suitable combination of the foregoing. The computer-readable storage media used herein are not to be construed as transient signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses through fiber optic cables), or electrical signals transmitted through wires.

[0146] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0147] Computer program instructions used to perform the operations described herein may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, status setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing status information from the computer-readable program instructions. This electronic circuitry can execute the computer-readable program instructions to implement various aspects of this specification.

[0148] Various aspects of this specification are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer-readable program instructions.

[0149] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that, when executed by the processor of the computer or other programmable data processing apparatus, they create means for implementing the functions / actions specified in one or more blocks of the flowchart and / or block diagram. These computer-readable program instructions can also be stored in a computer-readable storage medium that causes a computer, programmable data processing apparatus, and / or other device to operate in a particular manner; thus, the computer-readable medium storing the instructions comprises an article of manufacture that includes instructions for implementing aspects of the functions / actions specified in one or more blocks of the flowchart and / or block diagram.

[0150] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to perform the functions / actions specified in one or more boxes of a flowchart and / or block diagram.

[0151] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this specification. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of an instruction containing one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than those shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0152] The computer program product can be implemented specifically through hardware, software, or a combination thereof. In one alternative embodiment, the computer program product is specifically embodied in a computer storage medium; in another alternative embodiment, the computer program product is specifically embodied in a software product, such as a software development kit (SDK), etc.

[0153] Various embodiments of this specification have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical applications, or improvements to the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. An intelligent review system, characterized in that, The intelligent review system includes a cloud server, a terminal server, and a user terminal; the cloud server is used to: upon receiving a model deployment instruction, acquire a database containing multiple sample bid texts, and extract sample sub-texts from the sample bid texts according to multiple preset review contents; From the review model, a review module matching the review content represented by the sample sub-text is selected, and the review module is used to review the sample sub-text to obtain a sample sub-review value. Feature extraction is performed on the sample sub-text to obtain sample features, and based on the sample features, sample sub-texts representing the same review content are grouped to obtain multiple sample sub-text groups. Based on the sample sub-text groups, two sample sub-texts representing the same review content are selected, and the review model is adjusted according to the relationship between the sample sub-review values ​​of the two sample sub-texts before being sent to the terminal server. The relationship is a difference or ratio; the review model is adjusted when the difference or ratio between the sample sub-review values ​​of the two sample sub-texts exceeds a first preset value, or when the difference or ratio between the sample sub-review values ​​of the two sample sub-texts is less than a second preset value. The user terminal is used to send the tender text to be reviewed to the terminal server. The terminal server is used to review the bid text sent by the user terminal after receiving and deploying the review model sent by the cloud server, obtain the review result, and send the review result to the user terminal.

2. The system according to claim 1, characterized in that, When selecting two sample subtexts representing the same review content based on the sample subtext group, and adjusting the parameters of the review model according to the relationship between the sample sub-review values ​​of the two sample subtexts, the cloud server is used to: select two sample subtexts from the same sample subtext group, and obtain the review module that matches the review content represented by the sample subtexts; and adjust the parameters of the corresponding review module represented by the sample subtexts if the difference or ratio between the sample sub-review values ​​of the two sample subtexts exceeds a first preset value.

3. The system according to claim 1, characterized in that, When selecting two sample sub-texts representing the same review content based on the sample sub-text group, and adjusting the parameters of the review model according to the relationship between the sample sub-review values ​​of the two sample sub-texts, the cloud server is used to: obtain a first description text of the sample sub-text group, the first description text being used to describe the common features of the sample sub-texts in the sample sub-text group; obtain two target sample sub-text groups from multiple sample sub-text groups representing the same review content, wherein there are opposite description contents in the first description texts of the two target sample sub-text groups; select one sample sub-text from each of the two target sample sub-text groups, and adjust the parameters of the review module matched by the target sample sub-text group if the difference or ratio between the sample sub-review values ​​of the two sample sub-texts is less than a second preset value.

4. The system according to claim 3, characterized in that, When obtaining the first descriptive text of the sample sub-text group, the cloud server is configured to: use the sample sub-text group as text data of a preset large language model; use the review indicators contained in the review content represented by the sample sub-text as format guidance data of the large language model; input the text data and the format guidance data into the large language model, and use the text output by the large language model based on the text data and the format guidance data as the first descriptive text.

5. The system according to claim 3, characterized in that, When obtaining two target sample subtext groups from multiple sample subtext groups representing the same review content, the cloud server is used to: obtain the weight value of each review indicator in the review content, the weight value being used to indicate the importance of the review indicator in the review content, and taking the review indicator corresponding to the weight value greater than a third preset value as the target review indicator; and selecting two sample subtext groups from multiple sample subtext groups representing the same review content, the two sample subtext groups with opposite descriptions of the target review indicator in the first description text, as the target sample subtext groups.

6. The system according to claim 5, characterized in that, When obtaining the weight values ​​of each review indicator in the review content, the cloud server is used to: obtain the marker words corresponding to the review content and the importance scores of the marker words; Extract the key words from the second descriptive text of the evaluation indicators; The importance score of the review indicator is determined based on the frequency and importance score of the marker words appearing in the second description text. The importance scores of each evaluation indicator in the evaluation content are normalized to obtain the weight value of each evaluation indicator in the evaluation content.

7. The system according to claim 1, characterized in that, When reviewing the bid text sent by the user terminal to obtain a review result, the terminal server is configured to: upon receiving the bid text sent by the user terminal, extract sub-texts from the bid text based on multiple preset review contents; select a review module from the review model deployed on the cloud server that matches the review contents represented by the sub-text, and use the review module to review the sub-text to obtain a sub-review value; sum the sub-review values ​​corresponding to each sub-text in the bid text to obtain the review value of the bid text, and obtain the review result based on the review value.

8. An intelligent review method, characterized in that, For use on a cloud server, the method includes: upon receiving a model deployment instruction, acquiring a database containing multiple sample bid texts, and extracting sample sub-texts from the sample bid texts according to multiple preset review contents; selecting a review module from the review model that matches the review contents represented by the sample sub-texts, and using the review module to review the sample sub-texts to obtain sample sub-review values; extracting features from the sample sub-texts to obtain sample features, and grouping sample sub-texts representing the same review contents based on the sample features to obtain multiple sample sub-text groups; and selecting samples representing the same review contents from the sample sub-text groups. The evaluation model takes two sample sub-texts and, based on the relationship between their sample sub-evaluation values, adjusts its parameters before sending it to the terminal server. The terminal server, after receiving and deploying the evaluation model sent by the cloud server, evaluates the bid text sent by the user terminal, obtains the evaluation result, and sends the result to the user terminal. The relationship is a difference or ratio. If the difference or ratio between the sample sub-evaluation values ​​of the two sample sub-texts exceeds a first preset value, or if the difference or ratio is less than a second preset value, the evaluation model's parameters are adjusted.

9. An intelligent review method, characterized in that, For use with a terminal server, the method includes: after receiving and deploying a review model sent by a cloud server, reviewing the tender text sent by a user terminal to obtain a review result; the review model is obtained by the cloud server after adjusting the parameters of the review model based on the relationship between the sample sub-review values ​​of two sample sub-texts; the two sample sub-texts are obtained by the cloud server selecting sample sub-texts representing the same review content based on a sample sub-text group; the sample sub-text group is obtained by the cloud server grouping sample sub-texts representing the same review content based on sample features; the sample features are obtained by the cloud server performing feature extraction on the sample sub-texts; and the sample sub-review values ​​are obtained by the cloud server using the review model. The review module is a module selected by the cloud server from the review model that matches the review content represented by the sample sub-text. The sample sub-text is extracted by the cloud server from sample bidding text based on multiple preset review contents. The sample bidding text is text contained in the database obtained by the cloud server upon receiving a model deployment instruction. The relationship is a difference or ratio. If the difference or ratio between the sample sub-review values ​​of two sample sub-texts exceeds a first preset value, or if the difference or ratio between the sample sub-review values ​​of two sample sub-texts is less than a second preset value, the parameters of the review model are adjusted. The review result is then sent to the user terminal.

10. An electronic device, characterized in that, include: processor; A memory for storing processor-executable instructions; wherein the processor is configured to invoke the instructions stored in the memory to execute the method of claim 8 or 9.

Images