Evaluation copywriting generation method and device

Abstract

The embodiment of the invention provides an evaluation copywriting generation method and device, and the method comprises the steps: obtaining an attribute data set family, the attribute data set comprises an attribute data set of a plurality of objects, the plurality of objects comprise to-be-evaluated objects, and the attribute data set of any object comprises a plurality of key value pairs composed of attributes and attribute values; determining evaluation standards of various attributes according to distribution of attribute values of various attributes in the attribute data set family; constructing a model cue word according to evaluation standards of various attributes and the attribute data set of the to-be-evaluated object; and inputting the model cue word into an evaluation copywriting generation model, and obtaining an evaluation copywriting of the to-be-evaluated object output by the evaluation copywriting generation model. The method and the device are used for efficiently generating the accurate evaluation copywriting.

Description

Technical Field

[0001] This application relates to the field of natural language processing technology, and in particular to a method and apparatus for generating evaluation text. Background Technology

[0002] Evaluation copy refers to textual descriptions, analyses, and assessments of products, services, people, events, and other similar objects. Through evaluation copy, users can quickly understand the strengths and weaknesses of relevant objects, thereby making more rational and informed judgments in areas such as consumption decisions, cooperation choices, and cognitive formation. Therefore, evaluation copy plays a crucial role in information dissemination and value judgment.

[0003] Currently, evaluation texts are primarily written manually. This process requires writers to manually integrate and convert massive amounts of information from numerous data sources into text. This method is not only inefficient, but also prone to subjective bias due to manual operation, thus affecting the accuracy of the evaluation texts and making it difficult to ensure that evaluation texts are both efficiently produced and possess the necessary accuracy. Summary of the Invention

[0004] In view of this, embodiments of this application provide a method and apparatus for generating evaluation texts, which are used to efficiently generate accurate evaluation texts.

[0005] To achieve the above objectives, the technical solutions provided in this application are as follows:

[0006] In a first aspect, embodiments of this application provide a method for generating evaluation copy, including:

[0007] Obtain an attribute dataset family, wherein the attribute dataset includes attribute datasets of multiple objects, the multiple objects include the object to be evaluated, and the attribute dataset of any object includes multiple key-value pairs consisting of attributes and attribute values;

[0008] Based on the distribution of attribute values ​​of various attributes in the attribute dataset family, the evaluation criteria for various attributes are determined;

[0009] Model prompts are constructed based on the evaluation criteria of various attributes and the attribute data set of the object to be evaluated;

[0010] Input the model prompts into the evaluation copy generation model, and obtain the evaluation copy of the object to be evaluated output by the evaluation copy generation model.

[0011] As an optional implementation of this application, before determining the evaluation criteria for various attributes based on the distribution of attribute values ​​in the attribute data set, the method further includes:

[0012] For each object's attribute data set, determine whether there are mutually exclusive key-value pairs in the attribute data set. The mutually exclusive key-value pairs are key-value pairs with the same attribute but different attribute values.

[0013] If there are mutually exclusive key-value pairs in the attribute data set of the first object, the credibility of each mutually exclusive key-value pair is calculated based on the update time, occurrence frequency and source score, and the key-value pairs in the attribute data set of the first object are cleaned based on the credibility of each mutually exclusive key-value pair.

[0014] As an optional implementation of this application, the calculation of the credibility of each mutually exclusive key-value pair based on update time, frequency of occurrence, and source score includes:

[0015] Based on the update time of each mutually exclusive key-value pair, obtain the timeliness score of each mutually exclusive key-value pair;

[0016] Based on the frequency of occurrence of each mutually exclusive key-value pair, obtain the diffusion score of each mutually exclusive key-value pair;

[0017] Based on the source scores of each mutually exclusive key-value pair, obtain the reliability score of each mutually exclusive key-value pair;

[0018] The timeliness score, diffusion score, and reliability score of each mutually exclusive key-value pair are weighted and summed to obtain the credibility of the mutually exclusive key-value pair.

[0019] As an optional implementation of this application, the step of determining the evaluation criteria for various attributes based on the distribution of attribute values ​​in the attribute data set includes:

[0020] Obtain the first and second preset percentiles of the attribute values ​​of various attributes in the attribute data set;

[0021] Evaluation criteria for various attributes are constructed based on the first and second preset percentiles of the attribute values.

[0022] As an optional implementation of this application, the method further includes:

[0023] Periodically collect data to obtain a set of attribute data for at least one object;

[0024] The primary key value pair is used to match the attribute data set of the at least one object and each attribute data set in the attribute dataset family; the primary key value pair is a key-value pair used to uniquely identify the object to which it belongs.

[0025] If the first attribute data set in the attribute data set of the at least one object matches the second attribute data set in the attribute dataset family, then each key-value pair in the first attribute data set is added to the second attribute data set;

[0026] If the first attribute data set does not match any of the attribute data sets in the attribute dataset family, then the first attribute data set is added to the attribute data set.

[0027] As an optional implementation of this application, the model prompts include: role setting instructions and text structure setting instructions;

[0028] The role setting instruction is used to set the role of the evaluation copy generation model, and the copy structure setting instruction is used to set the structure of the evaluation copy output by the evaluation copy generation model.

[0029] As an optional implementation of this application, the method further includes:

[0030] The evaluation text of the object to be evaluated is displayed in the user interface;

[0031] Receive user editing operations on the evaluation text;

[0032] In response to the editing operation, the evaluation text is edited.

[0033] Secondly, embodiments of this application provide an evaluation copywriting generation apparatus, comprising:

[0034] An acquisition unit is used to acquire an attribute dataset family, wherein the attribute dataset includes attribute datasets of multiple objects, the multiple objects include objects to be evaluated, and the attribute dataset of any object includes multiple key-value pairs consisting of attributes and attribute values;

[0035] The processing unit is used to determine the evaluation criteria for various attributes based on the distribution of attribute values ​​of various attributes in the attribute dataset family.

[0036] The construction unit is used to construct model prompt words based on the evaluation criteria of various attributes and the attribute data set of the object to be evaluated;

[0037] The generation unit is used to input the model prompt words into the evaluation copy generation model and obtain the evaluation copy of the object to be evaluated output by the evaluation copy generation model.

[0038] As an optional implementation of this application, the processing unit is further configured to, before determining the evaluation criteria of various attributes based on the distribution of attribute values ​​of various attributes in the attribute data set, determine whether there are mutually exclusive key-value pairs in the attribute data set of each object, wherein the mutually exclusive key-value pairs are key-value pairs with the same attribute but different attribute values; if there are mutually exclusive key-value pairs in the attribute data set of the first object, then calculate the credibility of each mutually exclusive key-value pair based on the update time, occurrence frequency and source score, and clean the key-value pairs in the attribute data set of the first object based on the credibility of each mutually exclusive key-value pair.

[0039] As an optional implementation of this application, the processing unit is specifically configured to: obtain the timeliness score of each mutually exclusive key-value pair based on the update time of each mutually exclusive key-value pair; obtain the diffusion score of each mutually exclusive key-value pair based on the occurrence frequency of each mutually exclusive key-value pair; obtain the reliability score of each mutually exclusive key-value pair based on the source score of each mutually exclusive key-value pair; and perform a weighted summation of the timeliness score, diffusion score, and reliability score of each mutually exclusive key-value pair to obtain the credibility of the mutually exclusive key-value pair.

[0040] As an optional implementation of this application, the construction unit is specifically used to obtain the first preset percentile and the second preset percentile of the attribute values ​​of various attributes in the attribute data set; and to construct evaluation criteria for various attributes based on the first preset percentile and the second preset percentile of the attribute values ​​of various attributes.

[0041] As an optional implementation of this application, the acquisition unit is further configured to periodically collect data to obtain an attribute data set of at least one object; match the attribute data set of the at least one object with each attribute data set in the attribute dataset family based on primary key value pairs; the primary key value pairs are key-value pairs used to uniquely identify the object; if a first attribute data set in the attribute data set of the at least one object matches a second attribute data set in the attribute dataset family, then each key-value pair in the first attribute data set is added to the second attribute data set; if the first attribute data set does not match any of the attribute data sets in the attribute dataset family, then the first attribute data set is added to the attribute data set.

[0042] As an optional implementation of this application, the model prompts include: role setting instructions and text structure setting instructions;

[0043] The role setting instruction is used to set the role of the evaluation copy generation model, and the copy structure setting instruction is used to set the structure of the evaluation copy output by the evaluation copy generation model.

[0044] As an optional implementation of this application, the generation unit is further configured to display the evaluation text of the object to be evaluated in the user interface; receive the user's editing operation on the evaluation text; and edit the evaluation text in response to the editing operation.

[0045] Thirdly, this application provides an evaluation copy generation apparatus, including: a memory and a processor, wherein the memory is used to store a computer program and the processor is used to cause the evaluation copy generation apparatus to implement the evaluation copy generation method described in any of the above embodiments when executing the computer program.

[0046] Fourthly, embodiments of this application provide a computer-readable storage medium that, when executed by a computing device, causes the computing device to implement any of the above-described evaluation text generation methods.

[0047] Fifthly, embodiments of this application provide a computer program product that, when run on a computer, enables the computer to implement any of the above-described evaluation text generation methods.

[0048] The evaluation text generation method provided in this application first obtains an attribute dataset family that includes attribute data sets of multiple objects when generating evaluation text for an object to be evaluated. Then, based on the distribution of attribute values ​​of various attributes in the attribute dataset family, it determines the evaluation criteria for various attributes. Next, it constructs model prompt words based on the evaluation criteria for various attributes and the attribute data set of the object to be evaluated. Finally, it inputs the model prompt words into an evaluation text generation model and obtains the evaluation text for the object to be evaluated output by the evaluation text generation model. Since the evaluation text generation method provided in this application can obtain evaluation text by constructing model prompt words and inputting them into the evaluation text generation model, it can improve the efficiency of evaluation text generation compared to manually writing evaluation text. Furthermore, because the model prompt words are constructed based on the evaluation criteria for various attributes and the attribute data set of the object to be evaluated, and the evaluation criteria for various attributes are determined based on the distribution of attribute values ​​of various attributes in the attribute dataset family, this application can also ensure the accuracy of the generated evaluation text. In summary, this application can efficiently generate accurate evaluation text. Attached Figure Description

[0049] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0050] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the accompanying drawings that need to be called in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 One of the flowcharts for the evaluation text generation method provided in the embodiments of this application;

[0052] Figure 2 The second flowchart of the evaluation text generation method provided in the embodiments of this application;

[0053] Figure 3 The third flowchart of the evaluation text generation method provided in the embodiments of this application;

[0054] Figure 4 This is a schematic diagram of the structure of the evaluation text generation device provided in the embodiments of this application;

[0055] Figure 5 A schematic diagram of the hardware structure of the evaluation text generation device provided in the embodiments of this application. Detailed Implementation

[0056] To better understand the above-mentioned objectives, features, and advantages of this application, the solution of this application will be further described below. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0057] Many specific details are set forth in the following description in order to provide a full understanding of this application, but this application may also be implemented in other ways different from those described herein. Obviously, the embodiments in the specification are only some embodiments of this application, and not all embodiments.

[0058] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design solutions. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a specific manner. Furthermore, in the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0059] This application provides a method for generating evaluation text, which can be executed by a mobile phone, personal computer, or other evaluation text generation device. (See also...) Figure 1 As shown, the method for generating evaluation copy may include the following steps:

[0060] S11. Obtain the attribute dataset family.

[0061] The attribute data set includes attribute data sets of multiple objects, the multiple objects include the object to be evaluated, and the attribute data set of any object includes multiple key-value pairs consisting of attributes and attribute values.

[0062] A set family is a set whose elements are also sets. That is, a set family is a special type of set, special in that each element of the set family is also a set. For example, {{1, 2}, {3, 4}, {5, 6}} is a set family, where {1, 2}, {3, 4}, and {5, 6} are the elements of this set family, and these elements are themselves sets.

[0063] In some embodiments, obtaining an attribute dataset family includes: obtaining the attribute dataset family from an attribute database. Specifically, when the attribute database is a local database, obtaining the attribute dataset family from the attribute database includes: reading the attribute dataset family from the attribute database; when the attribute database is a cloud database, obtaining the attribute dataset family from the attribute database includes: sending a request message to the attribute database requesting to read the attribute dataset family, and receiving the attribute dataset family sent by the attribute database.

[0064] For example, when the object in this application embodiment is a land parcel, the attribute data set of the attribute dataset family may include key-value pairs consisting of attributes such as name, geographical location, administrative region, transfer area, building area, commercial facilities, schools, transportation facilities, factories, roads, future plans, and corresponding attribute values.

[0065] S12. Based on the distribution of attribute values ​​of various attributes in the attribute dataset family, determine the evaluation criteria for various attributes.

[0066] In some embodiments, the distribution indicators of attribute values ​​may include: the mean, median, mode, and specified percentile of the attribute values. The mean reflects the average level of attribute values; the median indicates the middle position of the data distribution; for data with extreme values, the median may be more representative of the central trend than the mean; the mode displays the most frequently occurring attribute value, revealing the central tendency of the data. Specified percentiles indicate the position and relative position of a specified data point within the data distribution.

[0067] In some embodiments, determining the evaluation criteria for various attributes based on the distribution of attribute values ​​of various attributes in the attribute dataset family includes: for each attribute, traversing the key-value pairs of the attribute data set of each object in the attribute dataset family, extracting the attribute value of the attribute, analyzing and statistically analyzing the extracted attribute values ​​to obtain the distribution of the attribute values ​​of the attribute, and finally determining the evaluation criteria of the attribute based on the distribution of the attribute values ​​of the attribute.

[0068] In some embodiments, the evaluation criteria for various attributes are determined based on the distribution of attribute values ​​in the attribute dataset, including:

[0069] Obtain the first and second preset percentiles of the attribute values ​​of various attributes in the attribute data set; construct evaluation criteria for various attributes based on the first and second preset percentiles of the attribute values ​​of various attributes.

[0070] A percentile is a statistical measure used to describe the position of data within a distribution, denoted as p%. For a given set of data, the p-th percentile is a value such that p% of the data are less than or equal to this value, and (100-p)% of the data are greater than or equal to this value. For example, the 25th percentile means that 25% of the data are less than or equal to this value.

[0071] For example, the first preset percentile can be 10%, and the second preset percentile can be 90%. The evaluation criteria constructed based on the first preset percentile and the second preset percentile are as follows: if it is higher than the first preset percentile, it is considered to be at a higher level; if it is lower than the second preset percentile, it is considered to be at a lower level; and if it is between the first preset percentile and the second preset percentile, it is considered to be at an average level.

[0072] S13. Construct model prompts based on the evaluation criteria of various attributes and the attribute data set of the object to be evaluated.

[0073] That is, model prompt words are constructed based on the evaluation criteria of various attributes and the key-value pairs in the attribute data set of the object to be evaluated.

[0074] Model prompts are key instructional text fragments that guide artificial intelligence models to generate specific content or perform specific tasks. They can consist of elements such as task descriptions, object feature information, evaluation criteria, and style and format requirements. The task description clearly informs the model of the type of task to be completed; in this embodiment, the character description is the evaluation text generated for the object to be evaluated. The object feature information is the set of attribute data for the object to be evaluated, and the evaluation criteria are the evaluation standards for various attributes.

[0075] In some embodiments, the model prompts include: role setting instructions and text structure setting instructions;

[0076] The role setting instruction is used to set the role of the evaluation copy generation model, and the copy structure setting instruction is used to set the structure of the evaluation copy output by the evaluation copy generation model.

[0077] For example, the role setting instruction could be: "You are an experienced land appraisal specialist," and the text structure setting instruction could be: "Adopt a multi-dimensional analytical structure. Begin by describing the macro-regional environment of the land parcel, then elaborate in the middle according to the key dimensions of natural conditions, planning potential, development advantages and challenges, and elaborate in detail on the land parcel's topography, soil geology, climate and sunlight characteristics in the natural conditions section, thereby constructing a systematic, comprehensive and highly practical land appraisal text."

[0078] S14. Input the model prompt words into the evaluation copy generation model, and obtain the evaluation copy of the object to be evaluated output by the evaluation copy generation model.

[0079] The evaluation text generation model in this application embodiment can be any model with natural language generation capabilities that can perform semantic understanding, logical analysis, and text construction based on specific input prompts. For example, the evaluation text generation model can be a Large Language Model, a GPT model, etc.

[0080] The evaluation text generation method provided in this application first obtains an attribute dataset family that includes attribute data sets of multiple objects when generating evaluation text for an object to be evaluated. Then, based on the distribution of attribute values ​​of various attributes in the attribute dataset family, it determines the evaluation criteria for various attributes. Next, it constructs model prompt words based on the evaluation criteria for various attributes and the attribute data set of the object to be evaluated. Finally, it inputs the model prompt words into an evaluation text generation model and obtains the evaluation text for the object to be evaluated output by the evaluation text generation model. Since the evaluation text generation method provided in this application can obtain evaluation text by constructing model prompt words and inputting them into the evaluation text generation model, it can improve the efficiency of evaluation text generation compared to manually writing evaluation text. Furthermore, because the model prompt words are constructed based on the evaluation criteria for various attributes and the attribute data set of the object to be evaluated, and the evaluation criteria for various attributes are determined based on the distribution of attribute values ​​of various attributes in the attribute dataset family, this application can also ensure the accuracy of the generated evaluation text. In summary, this application can efficiently generate accurate evaluation text.

[0081] Reference Figure 2As shown, in some embodiments, in Figure 1 Based on the illustrated embodiment, the evaluation text generation method provided in this application further includes: performing the following steps before determining the evaluation criteria for various attributes according to the distribution of attribute values ​​of various attributes in the attribute data set:

[0082] S21. For each object's attribute data set, determine whether there are mutually exclusive key-value pairs in the attribute data set.

[0083] The mutually exclusive key-value pairs are key-value pairs with the same attribute but different attribute values.

[0084] For example, the attribute data set of an object is: {Name: Plot 1, Greening rate: 29%, Plot ratio: 4.5, Distance from noise source: 280m}. Since there are no key-value pairs with the same attributes in the attribute data set of the object, it can be determined that there are no mutually exclusive key-value pairs in the attribute data set.

[0085] For example, the attribute data set of an object is: {Name: Plot 2, Greening rate: 23%, Greening rate: 25%, Plot ratio: 2.5, Distance from noise source: 550m}. Since the key-value pairs "Greening rate: 23%" and "Greening rate: 25%" in the attribute data set of the object are key-value pairs with the same attribute but different attribute values, it can be determined that there are mutually exclusive key-value pairs in the attribute data set of the object.

[0086] Based on step S21, if it is determined that there are mutually exclusive key-value pairs in the attribute data set of the first object, then the following steps are further executed:

[0087] S22. Calculate the credibility of each mutually exclusive key-value pair based on update time, frequency of occurrence, and source score.

[0088] The first object is any one of the plurality of objects. If there are mutually exclusive key-value pairs in the attribute data set of an object, the attribute data sets of the plurality of objects are combined into the attribute data set of the first object and the method provided in the embodiments of this application is executed.

[0089] In some embodiments, the confidence level of each mutually exclusive key-value pair is calculated based on update time, frequency of occurrence, and source score, including:

[0090] Step 22a: Based on the update time of each mutually exclusive key-value pair, obtain the timeliness score of each mutually exclusive key-value pair.

[0091] For example, the update time of a key-value pair can be the publication time of the article or news article that retrieves the key-value pair.

[0092] In the process of data cleaning and fusion, timeliness score is an important reference. When multiple mutually exclusive key-value pairs exist, the key-value pair with the higher timeliness score is more likely to reflect the current true attribute status of the object. For example, for the greening rate attribute of a land parcel, there may be multiple mutually exclusive key-value pairs. By comparing their timeliness scores, the latest greening rate can be selected as the standard value for that land parcel in relevant business systems or data analysis, so as to make a more accurate evaluation.

[0093] Step 22b: Based on the frequency of occurrence of each mutually exclusive key-value pair, obtain the diffusion score of each mutually exclusive key-value pair.

[0094] Among multiple mutually exclusive key-value pairs, if a key-value pair has a higher diffusion score, it indicates that it is more prevalent, and key-value pairs that appear more widely generally have higher credibility. For example, if there are multiple mutually exclusive key-value pairs for the distance of noise sources to a land parcel, and the key-value pair "noise source distance: 500m" appears 5 times, while "noise source distance: 300m" appears once, then the diffusion score of the key-value pair "noise source distance: 500m" can be considered to be higher than that of "noise source distance: 300m".

[0095] Step 22c: Based on the source scores of each mutually exclusive key-value pair, obtain the reliability score of each mutually exclusive key-value pair.

[0096] Source score is a quantitative assessment of the quality, authority, and credibility of a data source. It can be determined based on various factors, such as the data provider's reputation, professionalism, and the scientific rigor and accuracy of the data collection methods. For example, data from government statistical reports, due to their rigorous data collection processes and high authority, may be assigned a high source score; while data randomly posted by anonymous users on forums may have a lower source score. Reliability score, on the other hand, is a quantitative representation of the credibility of mutually exclusive key-value pairs after comprehensively considering the source score. It reflects the score each key-value pair receives based on the reliability of its source when multiple mutually exclusive key-value pairs exist (i.e., the same attribute has different values), and is used to determine the credibility of the key-value pair during data processing (such as data cleaning and fusion).

[0097] Step 22d: The timeliness score, diffusion score and reliability score of each mutually exclusive key-value pair are weighted and summed to obtain the credibility of the mutually exclusive key-value pair.

[0098] S23. Clean the key-value pairs in the attribute data set of the first object based on the credibility of each mutually exclusive key-value pair.

[0099] In some embodiments, the key-value pairs in the attribute data set of the first object are cleaned based on the confidence level of each mutually exclusive key-value pair, including: retaining only the mutually exclusive key-value pairs with the highest confidence level.

[0100] In other embodiments, the key-value pairs in the attribute data set of the first object are cleaned based on the confidence level of each mutually exclusive key-value pair, including: taking the mutually exclusive key-value pair with the highest confidence level as the standard attribute information of the first object, and ignoring other mutually exclusive key-value pairs.

[0101] Reference Figure 3 As shown, in some embodiments, in Figure 1 Based on the illustrated embodiments, the method provided in this application further includes:

[0102] S31. Periodically collect data to obtain a set of attribute data for at least one object.

[0103] Since data collection is performed periodically, the data collection process may occur before, after, or concurrently with the evaluation text generation process. Figure 3 The data collection process between China and Israel is executed before the evaluation text generation process.

[0104] In some embodiments, data collection can be performed first each time the evaluation text generation is triggered, and then the evaluation text generation can be executed.

[0105] In some embodiments, data can be automatically crawled from platforms such as news websites, professional industry forums, and government information release networks based on web crawling technology for data collection.

[0106] S32. Match the attribute data set of the at least one object and each attribute data set in the attribute dataset family based on the primary key value pair.

[0107] The primary key value pair is a key-value pair used to uniquely identify the object to which it belongs.

[0108] For example, when the object is a land parcel, the primary key value pair may include: a key-value pair representing the name of the land parcel and / or a key-value pair representing the location of the land parcel.

[0109] In step S32 above, if the first attribute data set in the attribute data set of the at least one object matches the second attribute data set in the attribute dataset family, then step S33 is executed as follows:

[0110] S33. Add each key-value pair in the first attribute data set to the second attribute data set.

[0111] For example, if the first attribute data set is: {Location: City A, District B, Road C, Area: 30,000 square meters, Greening rate: 25%, Plot ratio: 4.0}, and the second attribute data set is: {Location: City A, District B, Road C, Area: 35,000 square meters, Greening rate: 25%, Planned use: Mixed commercial and residential land}, then after adding each key-value pair from the first attribute data set to the second attribute data set, the second attribute data set becomes: {Location: City A, District B, Road C, Area: 30,000 square meters, Greening rate: 25%, Plot ratio: 4.0, Location: City A, District B, Road C, Area: 35,000 square meters, Greening rate: 25%, Planned use: Mixed commercial and residential land}.

[0112] In step S32 above, if the first attribute data set does not match any of the attribute data sets in the attribute dataset family, then step S34 is executed as follows:

[0113] S34. Add the first attribute data set to the attribute data set.

[0114] If the first attribute data set does not match any of the attribute data sets in the attribute dataset family, it means that the attribute dataset family does not include the attribute data set of the object corresponding to the first attribute data set. The first attribute data set is the attribute data set of the newly collected object. Therefore, the first attribute data set is directly added to the attribute data set without merging with other attribute data sets.

[0115] In some embodiments, after obtaining the evaluation text of the object to be evaluated output by the evaluation text generation model, the evaluation text generation method further includes: displaying the evaluation text of the object to be evaluated in a user interface, receiving editing operations from the user on the evaluation text; and editing the evaluation text in response to the editing operations.

[0116] For example, editing operations may include one or more of the following: text replacement, text deletion, text addition, paragraph adjustment, format adjustment, and saving.

[0117] Based on the same inventive concept, as an implementation of the above method, this application embodiment also provides an evaluation copywriting generation device. This embodiment corresponds to the aforementioned method embodiment. For ease of reading, this embodiment will not repeat the details of the aforementioned method embodiment one by one, but it should be clear that the evaluation copywriting generation device in this embodiment can correspondingly implement all the contents of the aforementioned method embodiment.

[0118] This application provides an evaluation copywriting generation device. Figure 4 This is a schematic diagram of the structure of the evaluation copy generation device, such as... Figure 4 As shown, the evaluation copy generation device 400 includes:

[0119] Acquisition unit 41 is used to acquire an attribute dataset family, wherein the attribute dataset includes attribute datasets of multiple objects, the multiple objects include objects to be evaluated, and the attribute dataset of any object includes multiple key-value pairs consisting of attributes and attribute values;

[0120] Processing unit 42 is used to determine the evaluation criteria for various attributes based on the distribution of attribute values ​​of various attributes in the attribute dataset family;

[0121] Construction unit 43 is used to construct model prompt words based on the evaluation criteria of various attributes and the attribute data set of the object to be evaluated;

[0122] The generation unit 44 is used to input the model prompt words into the evaluation copy generation model and obtain the evaluation copy of the object to be evaluated output by the evaluation copy generation model.

[0123] As an optional implementation of this application, the processing unit 42 is further configured to, before determining the evaluation criteria of various attributes based on the distribution of attribute values ​​of various attributes in the attribute data set, determine whether there are mutually exclusive key-value pairs in the attribute data set of each object, wherein the mutually exclusive key-value pairs are key-value pairs with the same attribute but different attribute values; if there are mutually exclusive key-value pairs in the attribute data set of the first object, calculate the credibility of each mutually exclusive key-value pair based on the update time, occurrence frequency and source score, and clean the key-value pairs in the attribute data set of the first object based on the credibility of each mutually exclusive key-value pair.

[0124] As an optional implementation of this application, the processing unit 42 is specifically used to obtain the timeliness score of each mutually exclusive key-value pair based on the update time of each mutually exclusive key-value pair; obtain the diffusion score of each mutually exclusive key-value pair based on the occurrence frequency of each mutually exclusive key-value pair; obtain the reliability score of each mutually exclusive key-value pair based on the source score of each mutually exclusive key-value pair; and perform a weighted summation of the timeliness score, diffusion score, and reliability score of each mutually exclusive key-value pair to obtain the credibility of the mutually exclusive key-value pair.

[0125] As an optional implementation of this application, the construction unit 43 is specifically used to obtain the first preset percentile and the second preset percentile of the attribute values ​​of various attributes in the attribute data set; and to construct evaluation criteria for various attributes based on the first preset percentile and the second preset percentile of the attribute values ​​of various attributes.

[0126] As an optional implementation of this application, the acquisition unit 41 is further configured to periodically collect data to obtain an attribute data set of at least one object; match the attribute data set of the at least one object with each attribute data set in the attribute dataset family based on primary key value pairs; the primary key value pairs are key-value pairs used to uniquely identify the object; if the first attribute data set in the attribute data set of the at least one object matches the second attribute data set in the attribute dataset family, then each key-value pair in the first attribute data set is added to the second attribute data set; if the first attribute data set does not match any of the attribute data sets in the attribute dataset family, then the first attribute data set is added to the attribute data set.

[0127] As an optional implementation of this application, the model prompts include: role setting instructions and text structure setting instructions;

[0128] The role setting instruction is used to set the role of the evaluation copy generation model, and the copy structure setting instruction is used to set the structure of the evaluation copy output by the evaluation copy generation model.

[0129] As an optional implementation of this application, the generation unit 44 is further configured to display the evaluation text of the object to be evaluated in the user interface; receive the user's editing operation on the evaluation text; and edit the evaluation text in response to the editing operation.

[0130] The evaluation text generation device provided in this application embodiment can execute the evaluation text generation method provided in any of the above embodiments. Its implementation principle and technical effect are similar, and will not be described again here.

[0131] Based on the same inventive concept, this application also provides an evaluation copy generation device. Figure 5 This is a schematic diagram of the structure of the evaluation copy generation device provided in the embodiments of this application, such as... Figure 5 As shown, the evaluation text generation device provided in this embodiment includes: a memory 501 and a processor 502. The memory 501 is used to store computer programs, and the processor 502 is used to execute any of the evaluation text generation methods provided in the above embodiment when executing the computer programs.

[0132] Based on the same inventive concept, embodiments of this application also provide a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the computing device to implement any of the evaluation text generation methods provided in the above embodiments.

[0133] Based on the same inventive concept, this application also provides a computer program product that, when run on a computer, enables the computing device to implement any of the evaluation text generation methods provided in the above embodiments.

[0134] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code.

[0135] It is understood that the memory in some embodiments of this application may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory may be random access memory (RAM), which serves as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate Synchronous DRAM (DDRSDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DRRAM). The memory of the systems and methods described in this application is intended to include, but is not limited to, these and any other suitable types of memory.

[0136] The processor may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by integrated logic circuits in the processor's hardware or by instructions in software form. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in some embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in some embodiments of this application can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory; the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method.

[0137] It is understood that the embodiments described in this application can be implemented using hardware, software, firmware, middleware, microcode, or a combination thereof. For hardware implementation, the processing unit can be implemented in 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), general-purpose processors, controllers, microcontrollers, microprocessors, other electronic units for performing the functions described in this application, or combinations thereof. For software implementation, the technology described in this application can be implemented through modules (e.g., procedures, functions, etc.) that perform the functions described in this application. Software code can be stored in memory and executed by a processor. The memory can be implemented in the processor or external to the processor.

[0138] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some or all of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A method for generating evaluation copy, characterized in that, include: Obtain an attribute dataset family, wherein the attribute dataset includes attribute datasets of multiple objects, the multiple objects include the object to be evaluated, and the attribute dataset of any object includes multiple key-value pairs consisting of attributes and attribute values; Based on the distribution of attribute values ​​of various attributes in the attribute dataset family, the evaluation criteria for various attributes are determined; Model prompts are constructed based on the evaluation criteria of various attributes and the attribute data set of the object to be evaluated; Input the model prompts into the evaluation copy generation model, and obtain the evaluation copy of the object to be evaluated output by the evaluation copy generation model.

2. The method according to claim 1, characterized in that, Before determining the evaluation criteria for each attribute based on the distribution of attribute values ​​in the attribute dataset, the method further includes: For each object's attribute data set, determine whether there are mutually exclusive key-value pairs in the attribute data set. The mutually exclusive key-value pairs are key-value pairs with the same attribute but different attribute values. If there are mutually exclusive key-value pairs in the attribute data set of the first object, the credibility of each mutually exclusive key-value pair is calculated based on the update time, occurrence frequency and source score, and the key-value pairs in the attribute data set of the first object are cleaned based on the credibility of each mutually exclusive key-value pair.

3. The method according to claim 2, characterized in that, The calculation of the credibility of each mutually exclusive key-value pair based on update time, frequency of occurrence, and source score includes: Based on the update time of each mutually exclusive key-value pair, obtain the timeliness score of each mutually exclusive key-value pair; Based on the frequency of occurrence of each mutually exclusive key-value pair, obtain the diffusion score of each mutually exclusive key-value pair; Based on the source scores of each mutually exclusive key-value pair, obtain the reliability score of each mutually exclusive key-value pair; The timeliness score, diffusion score, and reliability score of each mutually exclusive key-value pair are weighted and summed to obtain the credibility of the mutually exclusive key-value pair.

4. The method according to claim 2, characterized in that, The step of determining the evaluation criteria for various attributes based on the distribution of attribute values ​​in the attribute data set includes: Obtain the first and second preset percentiles of the attribute values ​​of various attributes in the attribute data set; Evaluation criteria for various attributes are constructed based on the first and second preset percentiles of the attribute values.

5. The method according to claim 1, characterized in that, The method further includes: Periodically collect data to obtain a set of attribute data for at least one object; The primary key value pair is used to match the attribute data set of the at least one object and each attribute data set in the attribute dataset family; the primary key value pair is a key-value pair used to uniquely identify the object to which it belongs. If the first attribute data set in the attribute data set of the at least one object matches the second attribute data set in the attribute dataset family, then each key-value pair in the first attribute data set is added to the second attribute data set; If the first attribute data set does not match any of the attribute data sets in the attribute dataset family, then the first attribute data set is added to the attribute data set.

6. The method according to claim 1, characterized in that, The model prompts include: character setting instructions and text structure setting instructions; The role setting instruction is used to set the role of the evaluation copy generation model, and the copy structure setting instruction is used to set the structure of the evaluation copy output by the evaluation copy generation model.

7. The method according to claim 1, characterized in that, After obtaining the evaluation text of the object to be evaluated output by the evaluation text generation model, the method further includes: The evaluation text of the object to be evaluated is displayed in the user interface; Receive user editing operations on the evaluation text; In response to the editing operation, the evaluation text is edited.

8. An evaluation copywriting generation device, characterized in that, include: A memory and a processor, wherein the memory is used to store a computer program and the processor is used to cause the evaluation copy generation apparatus to implement the evaluation copy generation method according to any one of claims 1-7 when executing the computer program.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a computing device, causes the computing device to implement the evaluation text generation method according to any one of claims 1-7.

10. A computer program product, characterized in that, When the computer program product is run on a computer, the computer enables the evaluation text generation method according to any one of claims 1-7.

Images