A corpus processing method, device, equipment and computer storage medium

Abstract

The application discloses a corpus processing method and device, equipment and a computer storage medium. A plurality of first quality evaluation index values of corpus data in a first time period are obtained respectively; the plurality of first quality evaluation index values are processed to obtain a plurality of second quality evaluation indexes; it is determined whether a target evaluation index meets a corresponding preset condition; and in response to the target quality evaluation index not meeting the corresponding preset condition, the reference priority of the corpus data is adjusted. According to the embodiment of the application, the rationality of the reference priority of the question and answer corpus can be improved, and a more reasonable reply statement is matched, thereby improving the user experience.

Description

Technical Field

[0001] This application belongs to the field of natural language processing technology, and in particular relates to a corpus processing method, apparatus, device and computer storage medium. Background Technology

[0002] Currently, natural language processing technology has been widely applied in the field of artificial intelligence, such as in smart supply chains, unmanned banks, smart branches, and smart cities. The application scenarios involved are also diverse, such as service robots, smart speakers, and intelligent customer service.

[0003] In typical intelligent question-and-answer application scenarios, the question-and-answer corpora involved are massive. Relying solely on manual configuration of citation priorities for these corpora is extremely labor-intensive and prone to inaccuracies. Existing solutions merely configure the citation priorities of the question-and-answer corpora once during storage using an algorithm, and then leave them unchanged after the corpora are put into use. However, over time, the parameters in the question-and-answer corpus database change, and the application scenarios suitable for the corpora also change. This reduces the rationality of the originally configured citation priorities, making it impossible to match accurate response statements based on citation priorities, resulting in a poor user experience. Summary of the Invention

[0004] This application provides a corpus data processing method, apparatus, device, and computer storage medium that can improve the rationality of the citation priority of question-and-answer corpora, thereby enabling the matching of more reasonable response statements and improving the user experience.

[0005] In a first aspect, embodiments of this application provide a corpus data processing method, the method comprising:

[0006] The first indicator values ​​of each of the multiple first quality evaluation indicators of the corpus data within the first time period are obtained respectively.

[0007] The values ​​of each of the multiple first quality evaluation indicators are processed to obtain multiple second quality evaluation indicators;

[0008] Determine whether the target evaluation indicators meet their corresponding preset conditions; the target evaluation indicators include at least one of multiple first quality evaluation indicators and multiple second quality evaluation indicators;

[0009] In response to the failure of the target quality evaluation indicators to meet their corresponding preset conditions, the citation priority of the corpus data is adjusted.

[0010] In some embodiments, the plurality of first quality evaluation indicators include application scenario-level indicators and corpus data hit rate indicators; the indicator values ​​of the plurality of first quality evaluation indicators are processed to obtain second quality evaluation indicators, including:

[0011] If the hit rate of the corpus data is less than the first preset threshold, the application scenario level will be adjusted.

[0012] The scope of corpus data referenced is determined based on the adjusted application scenario level;

[0013] Obtain the citation range index value of the corpus corresponding to the citation range, and obtain the second quality evaluation index of the application scenario level.

[0014] In some embodiments, determining whether a target evaluation indicator meets its corresponding preset conditions includes:

[0015] If the value of the corpus data hit rate indicator is less than the first preset threshold, it is determined that the application scenario level indicator does not meet the preset conditions.

[0016] In some embodiments, in response to a target quality evaluation indicator failing to meet its corresponding preset conditions, the citation priority of the corpus data is adjusted, including:

[0017] In response to the failure of the application scenario-level indicators to meet the preset conditions, the weighted sum of the values ​​of the corpus citation range indicator and other indicators among the multiple first quality evaluation indicators is performed using preset weight values ​​to obtain the weighted sum of the corpus data quality evaluation; the other indicators are those other than the corpus citation range indicator among the multiple first quality evaluation indicators.

[0018] Configure the reference priority corresponding to the weighted sum of quality evaluation values ​​as the reference priority of the corpus data.

[0019] In some embodiments, the index values ​​of each of the plurality of first quality evaluation indicators are processed to obtain a second quality evaluation indicator, including:

[0020] Obtain the second indicator value of the third quality evaluation indicator in the second time period; the third quality evaluation indicator is at least one of multiple first quality evaluation indicators; the second time period is before the first time period;

[0021] Based on the first and second index values ​​corresponding to the third quality evaluation index, determine the rate of change of the index value of the third quality evaluation index.

[0022] Based on the rate of change of the indicator value, the second quality evaluation indicator of the third quality evaluation indicator is obtained.

[0023] In some embodiments, determining whether a target evaluation indicator meets its corresponding preset conditions includes:

[0024] If the rate of change of the indicator value is greater than the second preset threshold, it is determined that the second quality evaluation indicator does not meet its corresponding preset conditions.

[0025] In some embodiments, the target first quality evaluation metric includes at least one of user feedback evaluation metrics and corpus data hit rate metrics.

[0026] In some embodiments, the plurality of first quality evaluation indicators include a corpus data citation frequency indicator, a corpus data hit rate indicator, and a popularity indicator; the values ​​of each of the plurality of first quality evaluation indicators are processed to obtain second quality evaluation indicators, including:

[0027] In response to the fact that the citation frequency index of the corpus data is greater than the third preset threshold and the hit rate index of the corpus data is less than the fourth preset threshold, a popularity tag is added to the corpus data to obtain the second quality evaluation index of the citation frequency index and the hit rate index of the corpus data.

[0028] In some embodiments, determining whether a target evaluation indicator meets its corresponding preset conditions includes:

[0029] If the value of the corpus data citation frequency index is greater than the third preset threshold and the value of the corpus data hit rate index is less than the fourth preset threshold, it is determined that the corpus data citation frequency index and the corpus data hit rate index do not meet their corresponding preset conditions.

[0030] In some embodiments, the corpus data includes a first query corpus and multiple first answer corpora of the first query corpus; after obtaining the first index values ​​of each of the multiple first quality evaluation indicators of the corpus data within a first time period, the method further includes:

[0031] Obtain the semantic attribute values ​​for each of the first answer corpora;

[0032] The semantic attribute values ​​of multiple first-answer corpora are compared to obtain semantic attribute difference index values.

[0033] When the semantic attribute difference index value is greater than the preset difference index value, an abnormal corpus prompt message is output.

[0034] In some embodiments, the plurality of first quality evaluation indicators include a corpus data ambiguity indicator; the index values ​​of the plurality of first quality evaluation indicators are processed to obtain a second quality evaluation indicator, including:

[0035] If the value of the ambiguity index of the corpus data exceeds the fifth preset threshold, the corpus data will be split.

[0036] Based on the split and processed corpus data, the updated index value of the corpus data ambiguity index is obtained, thus yielding the second quality evaluation index of the corpus data ambiguity index.

[0037] In some embodiments, determining whether a target evaluation indicator meets its corresponding preset conditions includes:

[0038] If the updated indicator value is less than the sixth preset threshold, it is determined that the second quality evaluation indicator of the application scenario level indicator does not meet the preset conditions.

[0039] In some embodiments, adjusting the citation priority of corpus data includes:

[0040] The weighted sum of the values ​​of multiple first quality evaluation indicators is obtained by using preset weight values ​​to obtain the weighted sum of the quality evaluation values ​​of the corpus data.

[0041] Configure the reference priority corresponding to the weighted sum of quality evaluation values ​​as the reference priority of the corpus data.

[0042] In some embodiments, the reference priority corresponding to the weighted sum of quality evaluation values ​​is configured as the reference priority of the corpus data, including:

[0043] The weighted sum of the quality evaluation values ​​is compared with multiple seventh preset thresholds; the ranges of the multiple seventh preset thresholds do not overlap.

[0044] In response to the weighted sum of the quality evaluation values ​​falling within the range of the target seventh preset threshold, the priority corresponding to the target seventh preset threshold is configured as the citation priority of the corpus data; the target seventh preset threshold is one of multiple seventh preset thresholds.

[0045] In some embodiments, the corpus data includes query corpus, answer corpus, and channel tags, and the method further includes:

[0046] Obtain the second query corpus, the second answer corpus, and the second channel tags of the target corpus;

[0047] Find the third query corpus in the corpus to which the target corpus data belongs, whose similarity to the second query corpus is greater than the eighth preset threshold;

[0048] Obtain the third-answer corpus and third-channel tags corresponding to the third-question corpus;

[0049] In response to the fact that the similarity between the second and third answer corpora is greater than the eighth preset threshold, and the second and third channel labels are the same, the corpus data corresponding to the second query corpus and the corpus data corresponding to the third query corpus are merged to obtain the merged corpus data.

[0050] In some embodiments, the corpus data further includes application scenario-level tags; after searching for a third query corpus in the corpus to which the target corpus data belongs, whose similarity to the second query corpus is greater than an eighth preset threshold, the method further includes:

[0051] Obtain the second application scenario level labels of the target corpus data and the third application scenario level labels corresponding to the third query corpus, respectively;

[0052] In response to the fact that the similarity between the second and third answer corpora is greater than the eighth preset threshold, and the scenario level of the second application scenario level label is greater than the scenario level of the third application scenario level label, the third query corpus is deleted and the third answer corpus is added as the answer corpus of the second query corpus.

[0053] Secondly, embodiments of this application provide a corpus data processing apparatus, the apparatus comprising:

[0054] The acquisition module is used to acquire the first indicator values ​​of each of the multiple first quality evaluation indicators of the corpus data within the first time period.

[0055] The processing module is used to process the index values ​​of multiple first quality evaluation indicators to obtain multiple second quality evaluation indicators.

[0056] The determination module is used to determine whether the target evaluation indicators meet their corresponding preset conditions; the target evaluation indicators include at least one of multiple first quality evaluation indicators and multiple second quality indicators;

[0057] The adjustment module is used to adjust the citation priority of corpus data in response to the failure of the target quality evaluation index to meet its corresponding preset conditions.

[0058] Thirdly, embodiments of this application provide a corpus data processing device, the device including: a processor and a memory storing computer program instructions;

[0059] When the processor executes computer program instructions, it implements a corpus data processing method as described in any of the first aspects.

[0060] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement the corpus data processing method as described in any of the first aspects.

[0061] Fifthly, embodiments of this application provide a computer program product, wherein instructions in the computer program product, when executed by a processor of an electronic device, cause the electronic device to perform a corpus data processing method as described in any of the first aspects.

[0062] The corpus data processing method, apparatus, device, and computer storage medium of this application embodiment can obtain second quality evaluation indicators by processing multiple first quality evaluation indicators of corpus data, and determine whether multiple second quality evaluation indicators meet preset conditions, that is, determine whether the question-and-answer corpus data meets preset quality requirements. If there are second quality evaluation indicators that do not meet the preset conditions, the citation priority of the corpus data is adjusted, thereby realizing the adjustment of the citation priority of corpus data according to changes in corpus quality, improving the rationality of the citation priority of question-and-answer corpus, and thus being able to match more reasonable response statements, thereby improving the user experience. Attached Figure Description

[0063] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0064] Figure 1 This is a flowchart illustrating a corpus data processing method provided in one embodiment of this application;

[0065] Figure 2 This is a flowchart illustrating a corpus data processing method provided in another embodiment of this application;

[0066] Figure 3 This is a flowchart illustrating a corpus data processing method provided in another embodiment of this application;

[0067] Figure 4 This is a flowchart illustrating a corpus data processing method provided in another embodiment of this application;

[0068] Figure 5 This is a flowchart illustrating a corpus data processing method provided in one embodiment of this application;

[0069] Figure 6 This is a flowchart illustrating another corpus data processing method provided in one embodiment of this application;

[0070] Figure 7 This is a flowchart illustrating another corpus data processing method provided in one embodiment of this application;

[0071] Figure 8 This is a schematic diagram of the structure of a corpus data processing apparatus provided in one embodiment of this application;

[0072] Figure 9 A schematic diagram of the hardware structure of the corpus data processing method provided in this application embodiment. Detailed Implementation

[0073] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.

[0074] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes the element.

[0075] Furthermore, the acquisition, storage, use, and processing of data in this application's technical solution all comply with relevant national laws and regulations.

[0076] Before introducing the specific implementation methods of this application, the technical terms used in describing the implementation methods of this application will be introduced first.

[0077] Corpus: refers to language materials, which are the content of linguistic research. A corpus is the basic unit that makes up a corpus; a collection of texts is a corpus.

[0078] Natural Language Processing (NLP) refers to the technology of using natural language, the language humans use for communication, to interact and communicate with machines. Through human processing of natural language, computers can read and understand it. Although NLP involves multi-dimensional operations such as speech, syntax, semantics, and pragmatics, its basic task, simply put, is to segment the corpus to be processed into semantically rich lexical units based on ontology dictionaries, word frequency statistics, and contextual semantic analysis.

[0079] Currently, Natural Language Processing (NLP) technology has been widely applied in the field of artificial intelligence, such as in smart supply chains, unmanned banks, smart branches, and smart cities. Its application scenarios are also diverse, including service robots, smart speakers, and intelligent customer service. NLP technology is driving the continuous development and breakthroughs in language intelligence. From simple rule-based methods to statistical methods, and now to deep learning neural network-based methods, NLP technology has become increasingly mature and has achieved great success in many fields. The reason deep learning has achieved such success in NLP, besides the innovation of deep learning algorithms, is primarily due to the massive amounts of linguistic data. If artificial intelligence is the engine, then data is the fuel, and linguistic data is the foundation of NLP applications.

[0080] Typically, the question-and-answer corpora involved in various intelligent question-and-answer application scenarios in NLP are massive. Relying solely on manual configuration of citation priorities for these corpora is extremely labor-intensive and yields poor accuracy. Existing solutions merely configure the citation priorities of the question-and-answer corpora once during storage using an algorithm, and then leave the citation priorities unchanged after the corpora are put into use. However, over time, the relevant parameters in the question-and-answer corpus database change, and the application scenarios suitable for the corpora also change. This leads to a decrease in the rationality of the originally configured citation priorities, making it impossible to match accurate response statements based on citation priorities, resulting in a poor user experience.

[0081] To address the aforementioned issues, this application proposes a corpus data processing method, apparatus, device, and computer storage medium. This method processes multiple first quality evaluation indicators of the corpus data to obtain second quality evaluation indicators. It then determines whether each of these second quality evaluation indicators meets preset conditions, i.e., whether the question-and-answer corpus data meets preset quality requirements. If any second quality evaluation indicator fails to meet the preset conditions, the citation priority of the corpus data is adjusted. This allows for adjustment of the citation priority of the corpus data based on changes in corpus quality, improving the rationality of the citation priority of the question-and-answer corpus and enabling the matching of more reasonable response statements, thereby enhancing the user experience.

[0082] The following section first introduces the corpus data processing method provided in the embodiments of this application.

[0083] Figure 1 A flowchart illustrating a corpus data processing method according to an embodiment of this application is shown. Figure 1 As shown, the corpus data processing method provided in this application embodiment includes the following steps: S101-S104.

[0084] S101: Obtain the first indicator values ​​of each of the multiple first quality evaluation indicators of the corpus data within the first time period.

[0085] As a possible example, the first time period mentioned above can be the current time period, that is, the first indicator value of each of the multiple first quality evaluation indicators can be the latest collected value; the first time period can also be a pre-specified time period, that is, to obtain the first indicator value of each of the multiple first quality evaluation indicators within the pre-specified time period.

[0086] The primary quality evaluation metric is used to evaluate the quality of the corpus. The primary quality evaluation metric may include any one or more of the following: application scenario level metric; user feedback evaluation metric; continuous repeated question metric; ambiguity metric of question and answer corpus data; citation scope metric of question and answer corpus data; hit rate metric of question and answer corpus data; citation frequency metric of question and answer corpus data; popularity metric; and human intervention rate metric.

[0087] S102: Process the values ​​of each of the multiple first quality evaluation indicators to obtain multiple second quality evaluation indicators.

[0088] It is understandable that the first quality evaluation index alone cannot fully reflect the quality of the corpus. By processing the first quality evaluation index, a second quality evaluation index can be obtained, thereby evaluating the quality of the corpus from more aspects.

[0089] As an example, the first quality evaluation indicator is one whose value can be directly obtained through data collection. Based on the first quality evaluation indicator, by further processing the values ​​of multiple first quality evaluation indicators, the value of the second quality evaluation indicator can be obtained. For example, based on the newly collected value of the user feedback evaluation indicator and its historical value, the rate of change of the user feedback evaluation indicator is calculated. This rate of change is the second quality evaluation indicator, and it reflects the change in users' evaluation of the corpus quality.

[0090] S103: Determine whether the target evaluation index meets its corresponding preset conditions.

[0091] In this embodiment of the application, the target evaluation index includes at least one of a plurality of first quality evaluation indexes and a plurality of second quality evaluation indexes.

[0092] S104: In response to the target quality evaluation indicator failing to meet its corresponding preset conditions, adjust the citation priority of the corpus data; in response to the target quality evaluation indicator meeting its corresponding preset conditions, stop processing the corpus data.

[0093] It's understandable that the citation priority of a corpus is positively correlated with its quality; that is, the higher the quality of the corpus, the higher its citation priority. However, over time, parameters in the question-and-answer corpus database have changed, and the suitable application scenarios for the corpus have also changed, leading to a decrease in the quality of the corpus data. This necessitates lowering the citation priority of the corpus data. Conversely, as the application scenarios change and the quality of the corpus data improves, its citation priority needs to be increased. Therefore, it is necessary to readjust the citation priority of the corpus data based on quality evaluation indicators.

[0094] As an example, it is determined whether each of the first and second quality evaluation indicators meets its corresponding preset conditions. If at least one quality evaluation indicator fails to meet its preset condition, it indicates that the quality of the corpus data does not meet the preset corpus quality conditions. The citation priority of the corpus data is then adjusted to improve the rationality of the citation priority, thereby matching a more accurate response statement based on the citation priority.

[0095] In this step, the values ​​of the first and second quality evaluation indicators can be processed to comprehensively determine the quality of the corpus data based on the processing results, and then the citation priority can be adjusted based on the processing results.

[0096] The corpus data processing method provided in this application's embodiments processes multiple first quality evaluation indicators of the corpus data to obtain second quality evaluation indicators. It then determines whether each of the multiple second quality evaluation indicators meets preset conditions, i.e., whether the question-and-answer corpus data meets preset quality requirements. If any second quality evaluation indicator fails to meet the preset conditions, the citation priority of the corpus data is adjusted. This achieves the adjustment of the citation priority of the corpus data based on changes in corpus quality, improving the rationality of the citation priority of the question-and-answer corpus, thereby enabling the matching of more reasonable response statements and enhancing the user experience.

[0097] It should be noted that the corpus hit rate refers to the number of times the target corpus meets the query conditions, that is, the number of times the target corpus is found to meet the query conditions. Due to the deterioration of corpus quality, the corpus hit rate may decrease, which in turn reduces the rationality of the corpus data's citation priority.

[0098] Therefore, to further improve the rationality of corpus data citation priority, another corpus data processing method is proposed. Please refer to [link to relevant documentation]. Figure 2 This is a flowchart of a corpus data processing method provided in another embodiment of this application. Figure 2 As shown, the data processing method for this corpus includes: S201-S207.

[0099] S201: Obtain the first indicator value of each of the multiple first quality evaluation indicators of the corpus data within the first time period.

[0100] In this embodiment, S201 is the same as S101 in the embodiment of this application. For the sake of brevity, it will not be described in detail here. For details, please refer to the description in the embodiment of this application.

[0101] S202: In response to the fact that the hit rate of the corpus data is less than the first preset threshold, the application scenario level is reduced.

[0102] In this embodiment of the application, multiple first quality evaluation indicators may include application scenario level indicators and corpus data hit rate indicators.

[0103] It should be noted that, in this embodiment, the application scenarios are divided into multiple levels according to different application scenarios of the corpus. Taking a financial institution as an example, the financial institution may include a head office, a first branch, and a second branch. The corpus applied to the head office is applicable to both the head office and the first and second branches, and its application scenario level is level one. The corpus applied to the first branch is applicable to both the first and second branches, and its application scenario level is level two. The corpus applied to the second branch is applicable only to the second branch, and its application scenario level is level three.

[0104] It is understandable that when the hit rate of corpus data decreases, its universality decreases, meaning that the range of scenarios in which it can be applied shrinks. Therefore, it is necessary to lower the application scenario level so that the application scenario level of the corpus data matches the actual situation.

[0105] As an example, in response to a corpus data hit rate metric value falling below a first preset threshold, the application scenario level of the corpus data is reduced. This reduction can be done by a single level, or by determining multiple reduction levels based on the difference between the corpus data hit rate metric value and the first preset threshold.

[0106] S203: Determine the scope of corpus data citation based on the adjusted application scenario level.

[0107] Understandably, as the application scenario level of the corpus data is reduced, its scope of reference also narrows. The scope of reference for the corpus data can be redefined based on the adjusted application scenario level.

[0108] As an example, based on lowering the adjusted application scenario level, the scope of corpus data citation is narrowed to the range corresponding to the application scenario level. For instance, before the application scenario level adjustment, the corpus data citation range included one main organization in the first-level application scenario, three first-level branch organizations in the second-level application scenario, and five second-level branch organizations in the third-level application scenario. After the application scenario level is lowered from first-level to second-level, the corpus data citation range includes one main organization in the first-level application scenario and three first-level branch organizations in the second-level application scenario.

[0109] S204: Obtain the corpus citation range index value corresponding to the citation range, and obtain the second quality evaluation index of the application scenario level index.

[0110] It is understandable that different citation ranges correspond to different corpus citation range index values, and the corpus citation range index values ​​can be preset.

[0111] As an example, the citation range index value corresponding to the citation range is obtained. This citation range index is the second quality evaluation index of the application scenario level index, and the citation range index value is the second quality evaluation index value of the application scenario level index.

[0112] S205: Determine whether the hit rate of the corpus data is less than the first preset threshold, thereby determining whether the application scenario level indicator meets the preset conditions, that is, whether the second quality evaluation indicator meets the preset conditions.

[0113] Understandably, if the value of the corpus data hit rate indicator is less than the first preset threshold, it means that the value of the corpus data hit rate indicator does not meet the preset requirements, causing the application scenario level indicator to no longer meet the preset conditions.

[0114] S206: In response to the failure of the application scenario level indicator to meet the preset conditions, the indicator values ​​of the corpus citation range indicator and other indicators among the multiple first quality evaluation indicators are weighted and summed using preset weight values ​​to obtain the weighted sum value of the corpus data quality evaluation.

[0115] In this embodiment of the application, the other indicators are those other than the corpus citation range indicator among a plurality of first quality evaluation indicators.

[0116] It should be noted that each primary quality evaluation indicator has a corresponding weight value, which can be preset according to actual needs.

[0117] As an example, in response to the failure of the application scenario level indicator to meet the preset conditions, the indicator values ​​of the above-mentioned corpus citation range indicator and other indicators among the multiple first quality evaluation indicators other than the application scenario level indicator are weighted and summed using preset weight values ​​to obtain the weighted sum value of the corpus data quality evaluation.

[0118] S207: Configure the citation priority corresponding to the weighted sum of quality evaluation values ​​as the citation priority of the corpus data.

[0119] It is understandable that different weighted sums of quality evaluation values ​​correspond to different citation priorities. The higher the weighted sum of quality evaluation values, the higher the quality of the corpus data, and therefore the higher the corpus citation priority.

[0120] As an example, a reference priority corresponding to the weighted sum of quality evaluation values ​​is determined, and this reference priority is configured as the reference priority of the corpus data.

[0121] The above describes a corpus data processing method provided in this application embodiment. This method adjusts the application scenario level of the corpus data by adjusting the index value of the corpus data hit rate index, and adjusts the citation scope index value of the corpus data according to the application scenario level, thereby adjusting the citation scope of the corpus data to a more reasonable range. Based on the adjusted citation scope and multiple first quality evaluation indicators, a weighted sum of quality evaluation values ​​for comprehensive evaluation of corpus quality is calculated, and the citation priority of the corpus data is adjusted to the citation priority corresponding to the weighted sum of quality evaluation values. This achieves the adjustment of citation priority according to the corpus hit rate index, improves the rationality of citation priority, and thus enables the matching of more reasonable response statements, improving the user experience.

[0122] It should be noted that, during the use of corpus data, relevant parameters or labels may be adjusted according to actual needs, resulting in changes in the values ​​of quality evaluation indicators. These changes indicate that the quality of the corpus has changed, and the citation priority of the corpus data needs to be adjusted in a timely manner according to these changes.

[0123] Therefore, to further adjust citation priorities in a timely manner based on changes in corpus data quality, another corpus data processing method is proposed. Please refer to [link to relevant documentation]. Figure 3 This is a flowchart of a corpus data processing method provided in another embodiment of this application. Figure 3 As shown, the data processing methods for this corpus include: S301-S306.

[0124] S301: Obtain the first indicator values ​​of each of the multiple first quality evaluation indicators of the corpus data within the first time period.

[0125] In this embodiment, S301 is the same as S101 in the embodiment of this application. For the sake of brevity, it will not be described in detail here. For details, please refer to the description in the embodiment of this application.

[0126] S302: Obtain the second indicator value of the third quality evaluation indicator in the second time period.

[0127] In this embodiment of the application, the third quality evaluation index can be at least one of a plurality of first quality evaluation indices.

[0128] In some embodiments of this application, the target first quality evaluation indicator may include at least one of user feedback evaluation indicator and corpus data hit rate indicator.

[0129] In some embodiments of this application, the second time period is prior to the first time period.

[0130] User feedback evaluation metrics can include scores for user feedback information, scores for intent recognition matching of questions in the question-and-answer corpus, and scores for users repeatedly asking the same question. User feedback evaluation metric values ​​can be obtained by summing any one or more of the above scores.

[0131] Since the second time period is prior to the first time period, the second indicator value in the second time period is equivalent to the historical indicator value of the third quality evaluation indicator.

[0132] S303: Based on the first and second index values ​​corresponding to the third quality evaluation index, determine the rate of change of the index value of the third quality evaluation index.

[0133] S304: Based on the rate of change of the index value, the second quality evaluation index of the third quality evaluation index is obtained.

[0134] Because users' actual needs have changed, or the quality of the corpus has changed, the value of the same quality evaluation indicator has also changed in different time periods. The magnitude of the change in corpus quality can be assessed based on the rate of change of the indicator value.

[0135] As an example, based on the first indicator value in the first time period and the second indicator value in the second time period, the rate of change of the indicator value of the third quality evaluation indicator is calculated, and the value of the rate of change of the indicator is the second quality evaluation indicator value of the third quality evaluation indicator.

[0136] S305: Determine whether the rate of change of the indicator value is greater than the second preset threshold, so as to determine whether the second quality evaluation indicator meets its corresponding preset conditions.

[0137] As an example, if the rate of change of an indicator value is greater than a second preset threshold, it indicates that the rate of change of the indicator value does not meet the preset conditions, and thus the quality of the corpus does not meet the preset conditions.

[0138] S306: If the second quality evaluation indicator fails to meet its corresponding preset conditions, adjust the citation priority of the corpus data. If the second quality evaluation indicator meets its corresponding preset conditions, stop processing the corpus data.

[0139] In this embodiment, S306 is the same as S104 in the embodiment of this application. For the sake of brevity, it will not be described in detail here. For details, please refer to the description in the embodiment of this application.

[0140] The above is a corpus data processing method provided in the embodiments of this application. This method obtains historical index values ​​of quality evaluation indicators, calculates the index value change rate of the quality evaluation indicators based on newly obtained index values ​​and historical index values, and can timely evaluate the changes in the quality of corpus data based on the index value change rate, thereby realizing timely adjustment of the citation priority of the corpus.

[0141] It should be noted that in the corpus data, there are cases where a corpus is identified as a candidate corpus many times, but is actually used a low number of times. That is, the citation rate is high but the hit rate is low. This situation causes the corpus data to frequently occupy the computing resources of the query, but it is not used because it does not meet the conditions, resulting in low corpus quality.

[0142] Therefore, to further improve the rationality of corpus data citation priority, another corpus data processing method is proposed. Please refer to [link to relevant documentation]. Figure 4 This is a flowchart of a corpus data processing method provided in another embodiment of this application. Figure 4 As shown, the data processing methods for this corpus include: S401-S404.

[0143] S401: Obtain the first indicator value of each of the multiple first quality evaluation indicators of the corpus data within the first time period.

[0144] For the sake of brevity, S401 in this embodiment and S101 in this application embodiment will not be described in detail here. For detailed information, please refer to the description in the embodiment of this application.

[0145] S402: Determine whether the value of the citation frequency index of the corpus data is greater than the third preset threshold, and whether the value of the hit rate index of the corpus data is less than the fourth preset threshold.

[0146] In this embodiment of the application, multiple first quality evaluation indicators may include corpus data citation frequency indicators, corpus data hit rate indicators, and popularity indicators.

[0147] Understandably, if the citation frequency index of the corpus data is greater than the third preset threshold and the hit rate index of the corpus data is less than the fourth preset threshold, it indicates that the corpus data is low-quality corpus data with a high citation rate but a low hit rate.

[0148] As an example, a popularity tag is added to low-quality corpus data with high citation rate but low hit rate. The popularity tag indicates that the corpus data has low hit popularity, and the popularity value corresponding to the popularity tag is the second quality evaluation index value of the corpus data citation frequency index and the corpus data hit rate index.

[0149] S403: In response to the fact that the value of the corpus data citation frequency indicator is greater than the third preset threshold and the value of the corpus data hit rate indicator is less than the fourth preset threshold, it is determined that the second quality evaluation indicators of the corpus data citation frequency indicator and the corpus data hit rate indicator do not meet their corresponding preset conditions. Add popularity tags to the corpus data to obtain the second quality evaluation indicators of the corpus data citation frequency indicator and the corpus data hit rate indicator.

[0150] As an example, if the value of the corpus data citation frequency index is greater than the third preset threshold and the value of the corpus data hit rate index is less than the fourth preset threshold, it indicates that the corpus data is low-quality corpus data with a high citation rate but a low hit rate. The quality of the corpus data does not meet the preset quality conditions. That is, the second quality evaluation index for determining the corpus data citation frequency index and the corpus data hit rate index does not meet their corresponding preset conditions.

[0151] In some embodiments of this application, processing of the corpus data is stopped in response to the fact that the index value of the corpus data citation frequency index is not greater than a third preset threshold, or the index value of the corpus data hit rate index is not less than a fourth preset threshold.

[0152] S404: In response to the failure of the target quality evaluation index to meet its corresponding preset conditions, adjust the citation priority of the corpus data.

[0153] For the sake of brevity, S404 in this embodiment and S104 in this application embodiment will not be described in detail here. For detailed information, please refer to the description in the embodiment of this application.

[0154] In this embodiment of the application, after adjusting the citation priority of the corpus data, the corpus data is put into use. After a period of time, the hit rate index value of the corpus data is re-acquired. If the hit rate index value of the corpus data is greater than the fourth preset threshold, the popularity tag can be deleted.

[0155] The above is a corpus data processing method provided in the embodiments of this application. The method evaluates the quality of corpus data through corpus data citation frequency index and corpus data hit rate index. If the citation frequency index value is greater than the preset threshold and the corpus data hit rate index is less than the preset threshold, the citation priority of the corpus data is reconfigured, thereby improving the rationality of the corpus data citation priority.

[0156] It should be noted that the corpus data includes the query corpus and multiple answer corpora of the query corpus. Since the multiple answer corpora are the answers to the same query corpus, the semantic synonym values ​​and sentiment values ​​of the multiple answer corpora corresponding to the same query corpus should be similar. If they are not similar, it indicates that there are incorrect answer corpora.

[0157] Therefore, to further improve the quality of the answer data in the corpus, another corpus data processing method is proposed. Please refer to [link to relevant documentation]. Figure 5 This is a flowchart of a corpus data processing method provided in one embodiment of this application. Figure 5 As shown, the data processing method for this corpus includes: S501-S510.

[0158] S501: Obtain the first indicator values ​​of each of the multiple first quality evaluation indicators of the corpus data within the first time period.

[0159] In this embodiment, S501 is the same as S101 in the embodiment of this application. For the sake of brevity, it will not be described in detail here. For detailed information, please refer to the description in the embodiment of this application.

[0160] S502: Obtain the semantic attribute values ​​of each first answer corpus respectively.

[0161] In the embodiments of this application, the semantic attribute values ​​of the corpus data can be calculated using rule-based statistical algorithms or bidirectional maximum matching algorithms.

[0162] In this embodiment of the application, the corpus data may include a first query corpus and multiple first answer corpora of the first query corpus.

[0163] The semantic attribute values ​​proposed in this application embodiment may include semantic synonym values ​​and sentiment values.

[0164] S503: Compare the semantic attribute values ​​of multiple first-answer corpora to obtain semantic attribute difference index values.

[0165] S504: Determine whether the semantic attribute difference index value is greater than the preset difference index value.

[0166] S505: If the semantic attribute difference index value is greater than the preset difference index value, output a corpus anomaly message. If the semantic attribute difference index value is not greater than the preset difference index value, stop processing the corpus data.

[0167] It is understandable that if the semantic attribute values ​​of multiple answer corpora in the same query corpus differ significantly, then the answer corpus corresponding to the semantic attribute value that is much higher than the average value is the answer corpus with the problem.

[0168] As an example, the semantic attribute values ​​of each first answer corpus are obtained separately, and the semantic attribute values ​​of multiple first answer corpora are calculated and compared. The semantic attribute difference index value is obtained based on the difference comparison result. If the semantic attribute difference index value is greater than the preset difference index value, it indicates that there is an incorrect answer corpus, and the corpus abnormality prompt message is output so that the user can process the incorrect answer corpus according to the prompt message.

[0169] S506: Determine whether the value of the ambiguity index in the corpus data is greater than the fifth preset threshold.

[0170] S507: In response to the ambiguity index value of the corpus data exceeding the fifth preset threshold, the corpus data is split. In response to the ambiguity index value of the corpus data not exceeding the fifth preset threshold, the processing of the corpus data is stopped.

[0171] In this embodiment of the application, the first quality evaluation index may include a corpus data ambiguity index.

[0172] S508: Obtain the updated index value of the corpus data ambiguity index based on the split corpus data, and obtain the second quality evaluation index of the corpus data ambiguity index.

[0173] In this embodiment of the application, the index value of the ambiguity index of the corpus data can be calculated by a rule-based statistical algorithm or a bidirectional maximum matching algorithm.

[0174] As an example, if the value of the corpus data ambiguity index exceeds the fifth preset threshold, it indicates that the corpus data is ambiguous and needs to be split. Splitting the corpus data reduces ambiguity. After splitting the corpus data, the updated corpus data ambiguity index value is obtained; this updated value is the second quality evaluation index value for the corpus data ambiguity index.

[0175] S509: Determine whether the updated indicator value is less than the sixth preset threshold.

[0176] S510: In response to the updated indicator value being less than the sixth preset threshold, it is determined that the second quality evaluation indicator of the application scenario level indicator does not meet the preset conditions, and the citation priority of the corpus data is adjusted. In response to the updated indicator value not being less than the sixth preset threshold, processing of the corpus data is stopped.

[0177] As an example, if the updated value of the ambiguity index of the corpus data is less than the sixth preset threshold, it indicates that the corpus data has been split, the quality of the split corpus data has been improved, and therefore its citation priority should be adjusted accordingly.

[0178] The above describes the corpus data processing method provided in this application embodiment. This method compares the semantic synonym values ​​and sentiment values ​​of multiple answer corpora corresponding to the same query corpus in the question-and-answer corpus. For semantic synonym values ​​or sentiment values ​​with significant differences, it outputs corpus abnormality prompts so that users can process abnormal corpora according to the prompts, thereby improving the quality of answer corpora in the corpus data. In addition, by splitting ambiguous corpora, the ambiguity of the corpus is reduced, and the citation priority is reconfigured based on the split corpora, thereby further improving the rationality of the citation priority.

[0179] If the target evaluation index does not meet its corresponding preset conditions, it is necessary to further determine how to adjust the citation priority of the corpus data. Since the citation priority is adjusted according to the changes in the quality of the corpus, the value of the citation priority can be determined by comprehensively considering the values ​​of multiple quality evaluation indicators, thereby further improving the rationality of the citation priority.

[0180] Therefore, to further improve the rationality of corpus data citation priority, another corpus data processing method is proposed. Please refer to [link to relevant documentation]. Figure 6 This is a flowchart of another corpus data processing method provided in an embodiment of this application. Figure 6 As shown, the data processing methods for this corpus include: S601-S606.

[0181] S601: Obtain the first indicator values ​​of each of the multiple first quality evaluation indicators of the corpus data within the first time period.

[0182] S602: Process the values ​​of each of the multiple first quality evaluation indicators to obtain multiple second quality evaluation indicators.

[0183] S603: Determine whether the target evaluation index meets its corresponding preset conditions.

[0184] In this embodiment of the application, the target evaluation index includes at least one of a plurality of first quality evaluation indexes and a plurality of second quality evaluation indexes.

[0185] In this embodiment, S601 to S603 are the same as S101 to S103 in the embodiments of this application. For the sake of brevity, they will not be described in detail here. For detailed information, please refer to the description in the embodiments of this application.

[0186] S604: In response to the target quality evaluation indicator failing to meet its corresponding preset condition, the values ​​of multiple first quality evaluation indicators are weighted and summed using preset weight values ​​to obtain a weighted sum of the quality evaluation values ​​of the corpus data. In response to the target quality evaluation indicator meeting its corresponding preset condition, processing of the corpus data is stopped.

[0187] S605: Compare the weighted sum of the quality assessment values ​​with multiple seventh preset thresholds.

[0188] In the embodiments of this application, the ranges of the multiple seventh preset thresholds do not overlap.

[0189] S606: In response to the weighted sum of the quality evaluation values ​​falling within the range of the seventh preset threshold, the priority corresponding to the seventh preset threshold is configured as the reference priority of the corpus data.

[0190] In this embodiment of the application, the target seventh preset threshold is one of a plurality of seventh preset thresholds.

[0191] For example, several primary quality evaluation indicators include user feedback evaluation indicators, repeated question evaluation indicators, question-and-answer corpus data ambiguity indicators, and question-and-answer corpus data citation scope indicators. The user feedback evaluation indicator value is x1, the repeated question evaluation indicator value is x2, the question-and-answer corpus data ambiguity indicator value is x3, and the question-and-answer corpus data citation scope indicator value is x4. The weight values ​​for the user feedback evaluation indicator are w1, the repeated question evaluation indicator value is w2, the question-and-answer corpus data ambiguity indicator value is w3, and the question-and-answer corpus data citation scope indicator value is w4. The weighted sum of the quality evaluation values ​​of the corpus data, y, is calculated using the following formula:

[0192] y=f(w1x1+w2x2+w3x3+w4x4+w5x5+b)=f(z)

[0193] z=w1x1+w2x2+w3x3+w4x4+w5x5+b

[0194]

[0195] Where b is the offset value.

[0196] For example, if x1 = 3, x2 = 6, x3 = 10, x4 = 4, x5 = 2, w1 = 3, w2 = 2, w3 = 1, w4 = 7, w5 = 3, b = 5, then z = 3x³ + 2x² + 10x¹ + 4x⁷ + 2x⁵ + 5 = 66, Z1 = 70, then y = 1.

[0197] The above is a corpus data processing method provided in the embodiments of this application. This method uses preset weight values ​​to perform weighted summation on the respective index values ​​of multiple first quality evaluation indicators to obtain a weighted summation value of the corpus data quality evaluation. The citation priority of the corpus data is determined based on the weighted summation value of the quality evaluation, thereby adjusting the citation priority based on multiple quality evaluation indicators and further improving the rationality of the citation priority.

[0198] It should be noted that, due to the massive amount of data in the corpus, the workload of maintenance and annotation is enormous. Some of the data is duplicated, which increases the unnecessary workload of corpus maintenance, resulting in low corpus reuse rate and wasting a lot of system storage resources. It also wastes system computing resources when retrieving corpus data.

[0199] Therefore, to further streamline the corpus data, another corpus data processing method is proposed. Please refer to [link to relevant documentation]. Figure 7 This is a flowchart of another corpus data processing method provided in one embodiment of this application. Figure 7 As shown, the corpus data processing method may also include: S701-S708.

[0200] S701: Obtain the second query corpus, the second answer corpus, and the second channel tags of the target corpus data.

[0201] In this embodiment of the application, the corpus data may include query corpus, answer corpus, and channel tags.

[0202] It should be noted that the channel tag is used to indicate the corpus to which the corpus data belongs. For example, if the channel tag of the corpus data is "Third Branch," it means that the corpus data is used in the corpus of the Third Branch.

[0203] As an example, we obtain the second query corpus, the second answer corpus, and the second channel labels of the question-and-answer target corpus data.

[0204] S702: Find the third query corpus in the corpus to which the target corpus data belongs, whose similarity to the second query corpus is greater than the eighth preset threshold.

[0205] As an example, the third query corpus in the corpus to which the target corpus data belongs, whose similarity to the second query corpus is greater than the eighth preset threshold, can be found by any one or more of the following methods: similarity matching algorithms include cosine similarity algorithm; Jaccard 2B similarity algorithm; Levenshtein B distance algorithm.

[0206] S703: Obtain the third answer corpus and third channel tags corresponding to the third query corpus.

[0207] S704: Determine whether the similarity between the second answer corpus and the third answer corpus is greater than the eighth preset threshold, and whether the second channel label and the third channel label are the same.

[0208] S705: In response to the similarity between the second and third answer corpora exceeding the eighth preset threshold, and the second and third channel labels being identical, the corpus data corresponding to the second and third query corpora are merged to obtain merged corpus data. In response to the similarity between the second and third answer corpora not exceeding the eighth preset threshold, or the second and third channel labels not being identical, the processing of the corpus data is stopped.

[0209] It should be noted that the answers to the same query corpus may vary slightly in different application scenarios, which makes the corpus more applicable. Therefore, the corpus data can only be merged when the similarity of the query corpus is high, the similarity of the corresponding answer corpus is also high, and the corresponding channel tags are the same.

[0210] The above merging process can involve deleting one answer corpus from the second query corpus and the third answer corpus, or deleting one answer corpus from the second answer corpus and the third answer corpus.

[0211] As an example, when the similarity between the second and third answer corpora is greater than the eighth preset threshold, and the second and third channel labels are the same, the corpus data corresponding to the second query corpus and the corpus data corresponding to the third query corpus are merged to obtain merged corpus data. Thus, by calculating the similarity of the corpus data, similar corpus data with the same query corpus, answer corpus, and channel label are merged, thereby achieving corpus data simplification.

[0212] As another implementation of this application, in order to further simplify the corpus data, the following steps may be included after S702:

[0213] S706: Obtain the second application scenario level labels of the target corpus data and the third application scenario level labels corresponding to the third query corpus, respectively.

[0214] It should be noted that, in this embodiment, the application scenarios are divided into multiple levels according to different application scenarios of the corpus. Taking a financial institution as an example, the financial institution may include a head office, a first branch, and a second branch. The corpus applied to the head office is applicable to both the head office and the first and second branches, and its application scenario level is level one. The corpus applied to the first branch is applicable to both the first and second branches, and its application scenario level is level two. The corpus applied to the second branch is applicable only to the second branch, and its application scenario level is level three.

[0215] When searching for corpora that meet the query criteria, the search should first be conducted in the corpus data at the application scenario level one, that is, from the most common corpus data. If no corpus data that meets the query criteria is found, the search should then be conducted in the corpus data at the application scenario level two, and so on, until corpus data that meets the query criteria is found.

[0216] S707: Determine whether the similarity between the second answer corpus and the third answer corpus is greater than the eighth preset threshold, and whether the scenario level of the second application scenario level label is greater than the scenario level of the third application scenario level label.

[0217] S708: In response to the fact that the similarity between the second answer corpus and the third answer corpus is greater than the eighth preset threshold, and the scenario level of the second application scenario level label is greater than the scenario level of the third application scenario level label, delete the third query corpus and add the third answer corpus as the answer corpus of the second query corpus.

[0218] In this embodiment of the application, in response to the fact that the similarity between the second answer corpus and the third answer corpus is not greater than an eighth preset threshold, or the scenario level of the second application scenario level label is not greater than the scenario level of the third application scenario level label, the processing of the corpus data is stopped.

[0219] It should be noted that for question-and-answer corpora with similar query corpora and their corresponding answer corpora, if their respective application scenario levels are different, it indicates that the corpora are duplicated and need to be simplified.

[0220] As a possible example, if the similarity between the second and third answer corpora is greater than the eighth preset threshold, and the scenario level of the second application scenario level label is greater than the scenario level of the third application scenario level label, the third query corpus is deleted, and the third answer corpus is added as the answer corpus of the second query corpus. This achieves the deletion of the query corpus corresponding to the low scenario level corpus data and the merging of its corresponding answer corpus into the high scenario level corpus data, thereby saving storage space and further simplifying the corpus data.

[0221] The above is a corpus data processing method provided in the embodiments of this application. This method merges similar corpus data with the same query corpus, answer corpus, and channel tag by calculating the similarity of the corpus data, thereby simplifying the corpus data. In addition, it merges question-and-answer corpus data with similar query corpus and their corresponding answer corpus, but with different application scenario levels. That is, it deletes the query corpus corresponding to the low scenario level corpus data and merges its corresponding answer corpus into the high scenario level corpus data, thereby saving storage space.

[0222] Based on the corpus data processing method provided in the above embodiments, combined with the appendix Figure 8 This application describes the specific implementation of the corpus data processing device provided in the embodiments.

[0223] See Figure 8 This is a schematic diagram of the structure of a corpus data processing device provided in one embodiment of this application. The training device 800 for the bearing fault diagnosis model includes:

[0224] The first acquisition module 801 is used to acquire the first indicator values ​​of each of the multiple first quality evaluation indicators of the corpus data within the first time period.

[0225] The processing module 802 is used to process the index values ​​of each of the multiple first quality evaluation indicators to obtain multiple second quality evaluation indicators.

[0226] The determination module 803 is used to determine whether the target evaluation index meets its corresponding preset conditions; the target evaluation index includes at least one of multiple first quality evaluation indicators and multiple second quality evaluation indicators;

[0227] The adjustment module 804 is used to adjust the citation priority of the corpus data in response to the fact that the target quality evaluation index does not meet its corresponding preset conditions.

[0228] The training device for the bearing fault diagnosis model provided in this application embodiment processes multiple first quality evaluation indicators of the corpus data to obtain second quality evaluation indicators. It then determines whether the multiple second quality evaluation indicators meet preset conditions, that is, whether the question-and-answer corpus data meets preset quality requirements. If there are second quality evaluation indicators that do not meet the preset conditions, the citation priority of the corpus data is adjusted. This realizes the adjustment of the citation priority of the corpus data according to the changes in corpus quality, improves the rationality of the citation priority of the question-and-answer corpus, and thus can match more reasonable response statements, thereby improving the user experience.

[0229] As one implementation of this application, in order to further improve the rationality of the corpus data citation priority, the above-mentioned processing module 802 may further include an adjustment submodule, a first determination submodule and a first acquisition submodule, the determination module 803 may further include a second determination submodule, and the adjustment module 804 may further include a first weighted summation submodule and a first configuration submodule.

[0230] The adjustment submodule is used to reduce the adjustment application scenario level when the value of the corpus data hit rate indicator is less than the first preset threshold.

[0231] The first determination submodule is used to determine the scope of reference for corpus data based on the adjusted application scenario level;

[0232] The first acquisition submodule is used to obtain the corpus citation range index value corresponding to the citation range, and to obtain the second quality evaluation index at the application scenario level.

[0233] The second determination submodule is used to determine that the application scenario level indicator does not meet the preset conditions in response to the corpus data hit rate indicator value being less than the first preset threshold.

[0234] The first weighted summation submodule is used to respond to the application scenario level indicator not meeting the preset conditions by using preset weight values ​​to perform weighted summation on the value of the corpus citation range indicator and the value of other indicators among multiple first quality evaluation indicators to obtain the weighted summation value of the corpus data quality evaluation; other indicators are those other than the corpus citation range indicator among multiple first quality evaluation indicators.

[0235] The first configuration submodule is used to configure the reference priority corresponding to the weighted sum of quality evaluation values ​​as the reference priority of the corpus data.

[0236] As one implementation of this application, in order to further adjust the citation priority in a timely manner according to changes in the quality of the corpus data, the above-mentioned processing module 802 may also include a second acquisition submodule, a third determination submodule and a fourth determination submodule, and the determination module 803 may also include a fifth determination submodule.

[0237] The second acquisition submodule is used to acquire the second indicator value of the third quality evaluation indicator in the second time period; the third quality evaluation indicator is at least one of multiple first quality evaluation indicators; the second time period is before the first time period.

[0238] The third determination submodule is used to determine the rate of change of the index value of the third quality evaluation index based on the first index value and the second index value corresponding to the third quality evaluation index.

[0239] The fourth determination submodule is used to obtain the second quality evaluation index of the third quality evaluation index based on the rate of change of the index value.

[0240] The fifth determination submodule is used to determine that the second quality evaluation indicator does not meet its corresponding preset conditions in response to the indicator value change rate being greater than the second preset threshold.

[0241] In this embodiment of the application, the target first quality evaluation index includes at least one of the user feedback evaluation index and the corpus data hit rate index.

[0242] As one implementation of this application, in order to further improve the rationality of the corpus data citation priority, the above processing module 802 may also include an adding submodule, and the determining module 803 may also include a sixth determining submodule.

[0243] A submodule is added to add a popularity tag to the corpus data in response to the corpus data citation frequency index value being greater than the third preset threshold and the corpus data hit rate index value being less than the fourth preset threshold, thereby obtaining the second quality evaluation index of the corpus data citation frequency index and the corpus data hit rate index.

[0244] The sixth determination submodule is used to determine that the second quality evaluation indicators of the corpus data citation frequency indicator and the corpus data hit rate indicator do not meet their corresponding preset conditions when the indicator value of the corpus data citation frequency indicator is greater than the third preset threshold and the indicator value of the corpus data hit rate indicator is less than the fourth preset threshold.

[0245] As one implementation of this application, in order to further improve the quality of the answer corpus in the corpus data, the above-mentioned device may further include a second acquisition module, a comparison module and an output module, the above-mentioned processing module 802 may further include a processing submodule, a seventh determination submodule and an eighth determination submodule, and the determination module 803 may further include a determination submodule.

[0246] The second acquisition module is used to acquire the semantic attribute values ​​of each first answer corpus respectively;

[0247] The comparison module is used to compare the semantic attribute values ​​of multiple first-answer corpora to obtain semantic attribute difference index values;

[0248] The output module is used to output corpus abnormality prompts when the semantic attribute difference index value is greater than the preset difference index value.

[0249] The processing submodule is used to split the corpus data in response to the ambiguity index value of the corpus data being greater than the fifth preset threshold.

[0250] The seventh determination submodule is used to obtain the updated index value of the corpus data ambiguity index based on the split corpus data, and to obtain the second quality evaluation index of the corpus data ambiguity index.

[0251] The eighth determination submodule is used to determine that the second quality evaluation indicator of the application scenario level indicator does not meet the preset conditions when the updated indicator value is less than the sixth preset threshold.

[0252] As one implementation of this application, in order to further improve the rationality of the corpus data citation priority, the above-mentioned adjustment module 804 may also include a weighted summation submodule and a configuration submodule.

[0253] The second weighted summation submodule is used to perform weighted summation on the respective index values ​​of multiple first quality evaluation indicators using preset weight values ​​to obtain the weighted summation value of the corpus data quality evaluation.

[0254] The second configuration submodule is used to configure the reference priority corresponding to the weighted sum of quality evaluation values ​​as the reference priority of the corpus data.

[0255] In this embodiment of the application, the configuration submodule is specifically used to: compare the weighted sum of the quality evaluation with multiple seventh preset thresholds; the ranges of the multiple seventh preset thresholds do not overlap; in response to the weighted sum of the quality evaluation falling within the range of the target seventh preset threshold, configure the priority corresponding to the target seventh preset threshold as the reference priority of the corpus data; the target seventh preset threshold is one of the multiple seventh preset thresholds.

[0256] As one implementation of this application, in order to further simplify the corpus data, the above-mentioned apparatus may further include:

[0257] The third acquisition module is used to acquire the second query corpus, the second answer corpus, and the second channel tags of the target corpus data;

[0258] The search module is used to find the third query corpus in the corpus to which the target corpus data belongs, whose similarity to the second query corpus is greater than the eighth preset threshold;

[0259] The fourth acquisition module is used to acquire the third answer corpus and the third channel tags corresponding to the third query corpus;

[0260] The merging module is used to merge the corpus data corresponding to the second query corpus and the corpus data corresponding to the third query corpus in response to the fact that the similarity between the second answer corpus and the third answer corpus is greater than the eighth preset threshold and the second channel label is the same as the third channel label, so as to obtain the merged corpus data.

[0261] The fifth acquisition module is used to acquire the second application scenario level labels of the target corpus data and the third application scenario level labels corresponding to the third query corpus, respectively.

[0262] The deletion module is used to delete the third query corpus and add the third answer corpus as the answer corpus of the second query corpus when the similarity between the second answer corpus and the third answer corpus is greater than the eighth preset threshold, and the scenario level of the second application scenario level label is greater than the scenario level of the third application scenario level label.

[0263] To address the problems of the prior art, embodiments of this application provide a corpus data processing method, apparatus, device, and computer storage medium. The corpus data processing method provided in this application embodiment will be described first below.

[0264] Figure 9 A schematic diagram of the hardware structure of the corpus data processing method provided in an embodiment of this application is shown.

[0265] The corpus data processing device may include a processor 901 and a memory 902 storing computer program instructions.

[0266] Specifically, the processor 901 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0267] Memory 902 may include mass storage for data or instructions. For example, and not limitingly, memory 902 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 902 may include removable or non-removable (or fixed) media. Where appropriate, memory 902 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 902 is non-volatile solid-state memory.

[0268] In a particular embodiment, memory 902 includes read-only memory (ROM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically rewritable ROM (EAROM), or flash memory, or a combination of two or more of these.

[0269] The processor 901 reads and executes computer program instructions stored in the memory 902 to implement any of the corpus data processing methods in the above embodiments.

[0270] In one example, the corpus data processing device may further include a communication interface 903 and a bus 910. For example, Figure 9 As shown, the processor 901, memory 902, and communication interface 903 are connected through bus 910 and complete communication with each other.

[0271] The communication interface 903 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0272] Bus 910 includes hardware, software, or both, that couples components of an online data traffic metering device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 910 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, this application contemplates any suitable bus or interconnect.

[0273] This corpus data processing device can execute the online data traffic billing method in this application embodiment based on currently blocked spam SMS messages and SMS messages reported by users, thereby achieving a combination of... Figure 1 and Figure 2 The online data traffic billing method and apparatus are described.

[0274] Furthermore, in conjunction with the online data traffic billing methods in the above embodiments, this application embodiment can provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when these computer program instructions are executed by a processor, they implement any of the online data traffic billing methods in the above embodiments.

[0275] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0276] The functional blocks shown in the above-described structural diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0277] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0278] The aspects of this application have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer 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 these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by dedicated hardware performing the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0279] The above description is merely a specific implementation of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A corpus data processing method, characterized in that, include: The first indicator values ​​of multiple first quality evaluation indicators of the corpus data within the first time period are obtained respectively. The multiple first quality evaluation indicators include application scenario level indicators and corpus data hit rate indicators. The application scenario level indicator value decreases as the corpus data hit rate indicator value decreases. The values ​​of each of the plurality of first quality evaluation indicators are processed to obtain a plurality of second quality evaluation indicators; The process of processing the values ​​of each of the plurality of first quality evaluation indicators to obtain a plurality of second quality evaluation indicators includes: In response to the fact that the hit rate of the corpus data is less than a first preset threshold, the application scenario level is reduced, and the universality of the corpus data decreases as the application scenario level is reduced. The scope of reference for the corpus data is determined based on the reduced application scenario level, and the scope of reference for the corpus data shrinks as the application scenario level decreases; Obtain the corpus citation range index value corresponding to the citation range, and obtain the second quality evaluation index of the application scenario level index; Determine whether the target evaluation index meets its corresponding preset conditions; the target evaluation index includes at least one of the plurality of first quality evaluation indicators and the plurality of second quality indicators; In response to the target evaluation index failing to meet its corresponding preset conditions, the citation priority of the corpus data is adjusted.

2. The corpus data processing method according to claim 1, characterized in that, The process of determining whether the target evaluation index meets its corresponding preset conditions includes: If the value of the corpus data hit rate indicator is less than a first preset threshold, it is determined that the application scenario level indicator does not meet the preset conditions.

3. The corpus data processing method according to claim 2, characterized in that, The step of adjusting the citation priority of the corpus data in response to the target evaluation index failing to meet its corresponding preset conditions includes: In response to the application scenario level indicator not meeting the preset conditions, the indicator values ​​of the corpus citation range indicator and other indicators among the plurality of first quality evaluation indicators are weighted and summed using preset weight values ​​to obtain the weighted sum value of the corpus data quality evaluation; the other indicators are those other than the corpus citation range indicator among the plurality of first quality evaluation indicators. Configure the reference priority corresponding to the weighted sum of the quality evaluation values ​​as the reference priority of the corpus data.

4. The corpus data processing method according to claim 1, characterized in that, The process of processing the values ​​of each of the plurality of first quality evaluation indicators to obtain a plurality of second quality evaluation indicators includes: Obtain the second indicator value of the third quality evaluation indicator in the second time period; the third quality evaluation indicator is at least one of the plurality of first quality evaluation indicators; the second time period is before the first time period; Based on the first indicator value and the second indicator value corresponding to the third quality evaluation indicator, determine the rate of change of the indicator value of the third quality evaluation indicator; Based on the rate of change of the index value, the second quality evaluation index of the third quality evaluation index is obtained.

5. The corpus data processing method according to claim 4, characterized in that, The process of determining whether the target evaluation index meets its corresponding preset conditions includes: In response to the rate of change of the indicator value being greater than a second preset threshold, it is determined that the second quality evaluation indicator does not meet its corresponding preset condition.

6. The corpus data processing method according to claim 4, characterized in that, The target evaluation metrics include at least one of user feedback evaluation metrics and corpus data hit rate metrics.

7. The corpus data processing method according to claim 1, characterized in that, The plurality of first quality evaluation indicators include a corpus data citation frequency indicator, a corpus data hit rate indicator, and a popularity indicator; the process of processing the values ​​of each of the plurality of first quality evaluation indicators yields a plurality of second quality evaluation indicators, including: In response to the fact that the index value of the corpus data citation frequency index is greater than a third preset threshold and the index value of the corpus data hit rate index is less than a fourth preset threshold, a popularity tag is added to the corpus data to obtain a second quality evaluation index of the corpus data citation frequency index and the corpus data hit rate index.

8. The corpus data processing method according to claim 7, characterized in that, The process of determining whether the target evaluation index meets its corresponding preset conditions includes: In response to the fact that the index value of the corpus data citation frequency index is greater than a third preset threshold and the index value of the corpus data hit rate index is less than a fourth preset threshold, it is determined that the second quality evaluation index of the corpus data citation frequency index and the corpus data hit rate index does not meet their corresponding preset conditions.

9. The corpus data processing method according to claim 1, characterized in that, The corpus data includes a first query corpus and multiple first answer corpora of the first query corpus; After obtaining the first index values ​​of each of the multiple first quality evaluation indicators of the corpus data within the first time period, the method further includes: Obtain the semantic attribute values ​​for each of the first answer corpora; The semantic attribute values ​​of the multiple first answer corpora are compared to obtain semantic attribute difference index values. In response to the semantic attribute difference index value being greater than the preset difference index value, an abnormal corpus prompt message is output.

10. The corpus data processing method according to any one of claims 1-9, characterized in that, The plurality of first quality evaluation indicators includes a corpus data ambiguity indicator; the processing of the respective indicator values ​​of the plurality of first quality evaluation indicators to obtain a plurality of second quality evaluation indicators includes: In response to the ambiguity index value of the corpus data being greater than a fifth preset threshold, the corpus data is split. Based on the split and processed corpus data, the updated index value of the corpus data ambiguity index is obtained, and the second quality evaluation index of the corpus data ambiguity index is obtained.

11. The corpus data processing method according to claim 10, characterized in that, The process of determining whether the target evaluation index meets its corresponding preset conditions includes: If the updated value of the ambiguity index of the corpus data is less than the sixth preset threshold, it is determined that the second quality evaluation index of the application scenario level index does not meet the preset conditions.

12. The corpus data processing method according to claim 1, characterized in that, The adjustment of the citation priority of the corpus data includes: The weighted sum of the values ​​of the multiple first quality evaluation indicators is obtained by using preset weight values ​​to calculate the quality evaluation weighted sum of the corpus data. Configure the reference priority corresponding to the weighted sum of the quality evaluation values ​​as the reference priority of the corpus data.

13. The corpus data processing method according to claim 3 or 12, characterized in that, The step of configuring the reference priority corresponding to the weighted sum of the quality evaluation values ​​as the reference priority of the corpus data includes: The weighted sum of the quality evaluation values ​​is compared with multiple seventh preset thresholds; the ranges of the multiple seventh preset thresholds do not overlap. In response to the weighted sum of the quality evaluation values ​​falling within the range of the target seventh preset threshold, the priority corresponding to the target seventh preset threshold is configured as the reference priority of the corpus data; the target seventh preset threshold is one of the plurality of seventh preset thresholds.

14. The corpus data processing method according to claim 1, characterized in that, The corpus data includes query corpus, answer corpus, and channel tags, and the method further includes: Obtain the second query corpus, the second answer corpus, and the second channel tags of the target corpus; Find a third query corpus in the corpus to which the target corpus data belongs that has a similarity greater than an eighth preset threshold with the second query corpus; Obtain the third answer corpus and the third channel tags corresponding to the third query corpus; In response to the fact that the similarity between the second answer corpus and the third answer corpus is greater than an eighth preset threshold, and the second channel label is the same as the third channel label, the corpus data corresponding to the second query corpus and the corpus data corresponding to the third query corpus are merged to obtain merged corpus data.

15. The corpus data processing method according to claim 14, characterized in that, The corpus data also includes application scenario-level tags; after finding a third query corpus in the corpus to which the target corpus data belongs that has a similarity greater than an eighth preset threshold with the second query corpus, the method further includes: The second application scenario level label of the target corpus data and the third application scenario level label corresponding to the third query corpus are obtained respectively; In response to the fact that the similarity between the second answer corpus and the third answer corpus is greater than an eighth preset threshold, and the scenario level of the second application scenario level label is greater than the scenario level of the third application scenario level label, the third query corpus is deleted, and the third answer corpus is added as the answer corpus of the second query corpus.

16. A corpus data processing device, characterized in that, The device includes: The acquisition module is used to acquire the first index values ​​of each of the multiple first quality evaluation indicators of the corpus data within the first time period. The multiple first quality evaluation indicators include application scenario level indicators and corpus data hit rate indicators. The application scenario level indicator value decreases as the corpus data hit rate indicator value decreases. The processing module is used to process the index values ​​of the plurality of first quality evaluation indicators to obtain a plurality of second quality evaluation indicators; The adjustment submodule is used to reduce the application scenario level in response to the corpus data hit rate index value being less than a first preset threshold. The universality of the corpus data decreases as the application scenario level decreases. The first determining submodule is used to determine the reference range of the corpus data based on the reduced application scenario level, wherein the reference range of the corpus data shrinks as the application scenario level decreases; The first acquisition submodule is used to acquire the corpus citation range index value corresponding to the citation range, and obtain the second quality evaluation index of the application scenario level index. A determination module is used to determine whether a target evaluation indicator meets its corresponding preset conditions; the target evaluation indicator includes at least one of the plurality of first quality evaluation indicators and the plurality of second quality indicators; The adjustment module is used to adjust the citation priority of the corpus data in response to the target evaluation index not meeting its corresponding preset conditions.

17. A corpus data processing device, characterized in that, The device includes: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, it implements the corpus data processing method as described in any one of claims 1-15.

18. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions, which, when executed by a processor, implement the corpus data processing method as described in any one of claims 1-15.

19. A computer program product, characterized in that, When the instructions in the computer program product are executed by the processor of the electronic device, the electronic device performs the corpus data processing method as described in any one of claims 1-15.

Images