Data classification and model training method and device, storage medium and program product

Abstract

The invention provides a data classification and model training method and device, a storage medium and a program product, and the method comprises the steps: marking to-be-processed data, namely interaction data in a logistics service, through a logistics big language model, namely a big language model applied to the logistics field, obtaining a label of the to-be-processed data, and then, carrying out the marking of the to-be-processed data; and classifying the to-be-processed data based on the label of the to-be-processed data to obtain a category of the to-be-processed data, the category including at least one of customer opinions, courier opinions and customer service opinions. Therefore, on the basis that the big language model has strong text understanding and generating capabilities, the label of the to-be-processed data can be more accurately determined through the logistics big language model, then the category of the to-be-processed data is determined based on the label, and classification and arrangement of the to-be-processed data can be effectively realized.

Description

Technical Field

[0001] This application relates to the field of data processing technology, specifically to a data classification and model training method, device, storage medium, and program product. Background Technology

[0002] In the logistics industry, couriers and customers frequently encounter various issues during delivery, requiring consultation or feedback, which generates a large amount of data. With hundreds of thousands of such audio and / or text data points generated daily—such as customer calls complaining about delivery services or couriers seeking advice on work-related problems—it is difficult to quickly and effectively categorize and organize this data.

[0003] To address this issue, existing technologies typically rely on rules or simple machine learning to automatically label these audio and / or text data. However, the performance of this method is limited by the complexity of the rules and the quality of the training data. If the rules are poorly formulated or the training data is of poor quality, it will be difficult to quickly and effectively classify and organize the data, and relevant personnel will find it difficult to effectively manage and utilize this data. Summary of the Invention

[0004] Based on the defects and shortcomings of the existing technology, this application proposes a data classification and model training method, device, storage medium and program product, which can label the data to be processed through the logistics big language model to obtain the label of the data to be processed, and classify the data to be processed based on the label to obtain the category of the data to be processed, thus solving the problem of difficulty in quickly and effectively classifying and organizing data.

[0005] According to a first aspect of the embodiments of this application, a data classification method is provided, including:

[0006] The logistics big language model is used to label the data to be processed, and the data to be processed is then categorized based on the labels to obtain the categories of the data to be processed. The logistics big language model is a big language model applied to the logistics field, the data to be processed is interactive data in logistics services, and the categories of the data to be processed include at least one of customer opinions, courier opinions, and customer service opinions.

[0007] According to a second aspect of the embodiments of this application, a model training method is provided, comprising:

[0008] Acquire logistics knowledge, and fine-tune the large language model based on the logistics knowledge to obtain a logistics large language model. The logistics large language model is a large language model applied to the logistics field. It is used to label the data to be processed to obtain the labels of the data to be processed, so as to classify the data to be processed based on the labels of the data to be processed and obtain the category of the data to be processed. The data to be processed is interactive data in logistics services.

[0009] According to a third aspect of the embodiments of this application, a data classification apparatus is provided, comprising:

[0010] The tagging module is used to tag the data to be processed using a logistics big data language model to obtain the tags of the data to be processed; the classification module is used to classify the data to be processed based on the tags to obtain the categories of the data to be processed. The logistics big data language model is a big data language model applied to the logistics field, the data to be processed is interactive data in logistics services, and the categories of the data to be processed include at least one of customer opinions, courier opinions, and customer service opinions.

[0011] According to a fourth aspect of the embodiments of this application, a model training apparatus is provided, comprising:

[0012] The module acquires logistics knowledge; the module trains the large language model based on the logistics knowledge to obtain a logistics large language model. The logistics large language model is a large language model applied to the logistics field, used to label the data to be processed to obtain labels for the data, facilitating the classification of the data based on these labels to determine its category. The data to be processed is interactive data from logistics services, and the category includes at least one of customer feedback, courier feedback, and customer service feedback.

[0013] According to a fifth aspect of the embodiments of this application, an electronic device is provided, including a memory and a processor;

[0014] The memory is connected to the processor and is used to store programs;

[0015] The processor is used to implement the data classification method as described in the first aspect or the model training method as described in the second aspect by running the program in the memory.

[0016] According to a sixth aspect of the embodiments of this application, a storage medium is provided, on which a computer program is stored, and when the computer program is run by a processor, it implements the data classification method as described in the first aspect or the model training method as described in the second aspect.

[0017] According to a seventh aspect of the present application, a computer program product is provided, the computer program product including computer program instructions, which, when executed by a processor, cause the processor to perform the data classification method as described in the first aspect or the model training method as described in the second aspect.

[0018] Among the aforementioned data classification methods, devices, storage media, and program products, the logistics big language model (i.e., a big language model applied to the logistics field) can be used to label the data to be processed, i.e., the interactive data in logistics services, to obtain tags for the data to be processed. Then, based on these tags, the data to be processed is classified to obtain its category, which includes at least one of customer feedback, courier feedback, or customer service feedback. Thus, based on the powerful text understanding and generation capabilities of the big language model, the logistics big language model can more accurately determine the tags for the data to be processed, and then determine the category based on these tags, effectively achieving the classification and organization of the data to be processed. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating a data classification method provided in an embodiment of this application.

[0021] Figure 2 This is a flowchart illustrating a model training method according to an embodiment of this application;

[0022] Figure 3 This is a schematic diagram illustrating an automatic labeling and classification process according to an embodiment of this application;

[0023] Figure 4 This is a schematic diagram of the structure of a data classification device proposed in an embodiment of this application;

[0024] Figure 5 This is a schematic diagram of the structure of a model training device proposed in an embodiment of this application;

[0025] Figure 6 This is a schematic diagram of the structure of an electronic device proposed in an embodiment of this application. Detailed Implementation

[0026] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0027] As described in the background section, in the logistics industry, couriers and customers frequently encounter various problems during delivery, requiring consultation or feedback, which generates a large amount of data. Due to the sheer volume of data, classifying and organizing it is quite difficult. Existing technologies typically rely on rules or simple machine learning to label and categorize this data. However, the performance of these methods is limited by the complexity of the rules and the quality of the training data. If the rules are poorly formulated or the training data is of low quality, it becomes difficult to quickly and effectively classify and organize the data, making it difficult for relevant personnel to effectively manage and utilize this data to improve logistics services.

[0028] Building upon this foundation, the inventors further discovered that large language models, such as GPT, trained on massive amounts of text data, can learn rich linguistic knowledge and provide powerful text understanding and generation capabilities. Therefore, large language models can serve as core components for data labeling and classification. Labeling data using large language models can effectively improve the accuracy of the labels and can also handle more complex text data such as long texts and multilingual texts. Thus, after acquiring logistics knowledge and fine-tuning the large language model based on this knowledge to obtain a logistics large language model, the data to be processed—i.e., the interactive data in logistics services—can be labeled using this model to obtain accurate labels for the data. Subsequently, based on these labels, the data can be classified to obtain accurate categories, enabling rapid and effective classification and organization of the data. This allows relevant personnel to understand and improve logistics services based on the processed data.

[0029] Based on the above concept, this specification provides a data classification and model training method, which will be described exemplarily below with reference to the accompanying drawings.

[0030] Exemplary methods

[0031] Please see Figure 1In one exemplary embodiment, a data classification method is provided, applied to any electronic device, which can acquire data to be processed and perform model invocation. The data to be processed is interactive data in logistics services, such as customer service-related dialogue data, courier-related dialogue data, and customer-related dialogue data, wherein the dialogue data is text and / or voice data.

[0032] like Figure 1 As shown, the data classification method includes steps S101-S102:

[0033] S101: Using the logistics big language model, the data to be processed is labeled to obtain the labels of the data to be processed.

[0034] Among them, the logistics big language model is a big language model applied to the logistics field.

[0035] The logistics big language model is obtained by fine-tuning and training large language models such as GPT and llama based on logistics knowledge. Since large language models have the ability to understand semantics, adding a large amount of logistics knowledge to large language models can yield a logistics big language model, which has a stronger understanding ability in the logistics field.

[0036] By tagging the data to be processed using the logistics big data language model, and obtaining the tags for the data, the data can be directly input into the model for analysis. This analysis will yield a summary or abstract of the data, which will then be used as the tag for the data. Alternatively, the logistics big data language model can be combined with tags from a tag database to tag the data and obtain its own tags.

[0037] For example, the data to be processed is a voice recording of a courier saying, "Why haven't the company uniforms arrived yet?" By tagging the data to be processed using the logistics big data language model, the tag for the data to be processed can be determined as "unclear delivery time of courier uniforms".

[0038] The number of labels in the data to be processed is at least one.

[0039] For example, the data to be processed is a voice recording of a courier saying, "Why haven't the uniforms the company sent arrived yet? And the last uniforms were of such poor quality." By tagging this data using the logistics big data language model, the tags for the data to be processed can be determined as "unclear delivery time of courier uniforms" and "courier reports poor uniform quality."

[0040] S102: Based on the labels of the data to be processed, classify the data to be processed to obtain the category of the data to be processed.

[0041] The categories of data to be processed include at least one of the following: customer feedback, courier feedback, and customer service feedback.

[0042] For example, the data to be processed is labeled as "unclear delivery time of courier uniforms". Based on this label, the data to be processed can be classified as courier opinions.

[0043] Specifically, the opinions include those on timeliness and packaging of express parcels.

[0044] Understandably, the above opinions include at least one of the following: complaints, suggestions, or feedback.

[0045] In this embodiment, a large-scale language model for logistics (i.e., a large-scale language model applied to the logistics field) is used to label the data to be processed, i.e., the interactive data in logistics services, to obtain tags for the data to be processed. Then, based on these tags, the data to be processed is categorized to obtain its class, which includes at least one of customer feedback, courier feedback, and customer service feedback. Thus, leveraging the powerful text understanding and generation capabilities of the large-scale language model, the logistics large-scale language model can more accurately determine the tags for the data to be processed. Subsequently, based on these tags, the class of the data to be processed can be determined, effectively achieving the categorization and organization of the data to be processed.

[0046] To ensure that the label of each piece of data to be processed can be determined, in some embodiments, the data to be processed is tagged using a logistics big data language model. When the label of the data to be processed is obtained, the label database can be used to determine whether there is a label that matches the data to be processed.

[0047] The logistics big data language model is used to match the tags in the tag database with the data to be processed to obtain the matching results. Then, based on the matching results, the tags of the data to be processed are determined.

[0048] The matching results include tags in the tag database that match the data to be processed. In other words, the matching results indicate whether the tags in the tag database match the data to be processed, and which specific tags match the data to be processed.

[0049] If the matching result is empty, it means that none of the tags in the tag database match the data to be processed, or in other words, there are no tags in the tag database that match the data to be processed; if the matching result is not empty, it means that there are tags in the tag database that match the data to be processed.

[0050] When determining the labels for the data to be processed based on the matching results, if the matching results are not empty, the labels in the label database that match the content of the data to be processed are determined as the labels for the data to be processed; if the matching results are empty, the data to be processed is labeled using the labeling model to obtain the labels for the data to be processed.

[0051] That is, if the matching result is not empty, the label of the data to be processed is determined based on the label in the matching result that matches the data to be processed.

[0052] Specifically, the tags in the matching results that match the data to be processed are directly identified as the tags of the data to be processed.

[0053] For example, taking the tag "customer dissatisfied with timeliness" in the tag database as an example, the large language model determines whether the current corpus, i.e., the data to be processed, contains content with the tag "customer dissatisfied with timeliness". If it does, the tag "customer dissatisfied with timeliness" is determined to match the data to be processed; if it does not, the tag "customer dissatisfied with timeliness" is determined to not match the data to be processed. Then, the tag "customer dissatisfied with timeliness" is used as the tag for the data to be processed.

[0054] In addition, the aforementioned labeling model is used to label the data to be processed. This labeling model can be a logistics big data language model, or a model obtained by training a logistics big data language model based on the labeling task.

[0055] In this embodiment, due to the powerful text understanding and generation capabilities of the logistics big data language model, the tags in the tag database are matched with the data to be processed using the logistics big data language model, which can quickly and accurately obtain matching results. The matching results include tags in the tag database that match the data to be processed. Subsequently, based on the matching results, when determining the tags of the data to be processed, if the matching result is not empty, the tags in the tag database that match the content of the data to be processed are determined as the tags of the data to be processed, which can achieve the effect of quickly and accurately determining the tags of the data to be processed. If the matching result is empty, the data to be processed is tagged using a tagging model, which can accurately determine the tags of the data to be processed, effectively ensuring that the tag can be determined for each piece of data to be processed.

[0056] To ensure the rapid and accurate determination of the labels of the data to be processed, in some embodiments, the data to be processed is labeled using a labeling model. After obtaining the labels of the data to be processed, the labels of the data to be processed are stored in a label database.

[0057] In this way, when subsequent data containing the same content appears, the label of the data to be processed can be directly determined based on the labels in the label database, achieving the effect of quickly and accurately determining the label of the data to be processed.

[0058] Since the tags of the data to be processed are a summary and distillation of the content of the data, the tag classification of the data to be processed can represent the category of the data. In order to achieve the classification and organization of the data to be processed, in some embodiments, the data to be processed can be classified based on the tag classification corresponding to the tags of the data to be processed, so as to obtain the category of the data to be processed.

[0059] Specifically, based on the labels of the data to be processed, the data to be processed is classified. When obtaining the category of the data to be processed, the label category corresponding to the label of the data to be processed is first determined, and then the label category corresponding to the label of the data to be processed is determined as the category of the data to be processed.

[0060] When determining the label category corresponding to the labels of the data to be processed, the label category can be determined based on the correspondence between labels and label categories. Alternatively, a multimodal model can be used to analyze the labels of the data to be processed to determine the corresponding label categories.

[0061] The label categories include at least one of the following: customer feedback, courier feedback, and customer service feedback.

[0062] Similar to the categories of data to be processed, opinions in the label classification can be further divided into timeliness opinions, express packaging opinions, etc., based on different actual needs. Specifically, opinions can be complaints, suggestions, or feedback.

[0063] For example, considering data 1 and data 2 in the data to be processed, data 1 is an audio message stating, "You promised the package would be delivered yesterday, why hasn't it been delivered yet?", and data 2 is text containing the message, "You promised the package would be delivered yesterday, but it's only now being delivered, and the packaging is damaged." Through the aforementioned tagging process, the tag for data 1 is determined to be "Customer dissatisfied with delivery timeliness," and the tags for data 2 are "Customer dissatisfied with delivery timeliness" and "Damaged packaging." Then, based on the tags for data 1 and data 2, the tag category corresponding to the tag for data 1 is determined to be customer timeliness complaint, and the tag category corresponding to the tag for data 2 includes both customer timeliness complaint and packaging complaint. Therefore, the category for data 1 can be determined as customer timeliness complaint, and the category for data 2 can be determined as both customer timeliness complaint and packaging complaint.

[0064] Thus, in this embodiment, by determining the tag category corresponding to the tag of the data to be processed, and then classifying the tags corresponding to the tags of the data to be processed, the category of the data to be processed can be determined quickly and accurately, thereby realizing the classification and organization of the data to be processed and ensuring the accuracy of the classification and organization.

[0065] In order to achieve effective management and utilization of data, in some embodiments, the tag categories corresponding to all tags in the tag database are statistically analyzed and displayed.

[0066] The display method can be text, image, table, etc.

[0067] Of course, depending on the actual needs, in addition to displaying the tag categories corresponding to all tags in the database, the data to be processed under each tag in the tag database will also be displayed.

[0068] This allows relevant personnel to understand customer, courier, and customer service feedback on logistics services, enabling them to take timely measures to improve logistics services and achieve effective management and utilization of data. For example, when there are many customer complaints about delivery timeliness, relevant personnel can use this information to take timely measures to reduce the complaint rate and improve the level of logistics services.

[0069] Please see Figure 2 In one exemplary embodiment, a model training method is provided, which is applied to any electronic device that can acquire logistics knowledge and communicate with the electronic device to which the above-described data classification method is applied.

[0070] like Figure 2 As shown, the model training method includes steps S201-S202:

[0071] S201: Acquiring logistics knowledge.

[0072] S202: Fine-tune the large language model based on logistics knowledge to obtain a logistics large language model.

[0073] Among them, the logistics big language model is a big language model applied in the logistics field. It is used to label the data to be processed to obtain the labels of the data to be processed, so as to classify the data to be processed based on the labels and obtain the category of the data to be processed.

[0074] In addition, the data to be processed is interactive data in logistics services, and the categories of the data to be processed include at least one of customer opinions, courier opinions, and customer service opinions.

[0075] Thus, in this embodiment, by fine-tuning the large language model based on the acquired logistics knowledge, a large logistics language model that fully understands the knowledge in the logistics field can be obtained. Based on this large logistics language model, interactive data in logistics services can be identified and processed more accurately. Through this large logistics language model, the category of the data to be processed can be quickly and accurately identified.

[0076] In one exemplary embodiment, a model training method is provided, which is applied to any electronic device that can acquire corpus and its labels and can communicate with the electronic device to which the above data classification method is applied.

[0077] The model training method is described below, including steps S301-S302:

[0078] S301: Obtain the corpus and its labels.

[0079] The corpus consists of unprocessed data from historical time periods prior to the current moment, which is the interactive data in the logistics service.

[0080] S302: Fine-tune the large language model based on the corpus and its labels to obtain the labeling model.

[0081] Among them, the large language model can be a general model or the logistics large language model mentioned above.

[0082] The labeling model is used to label the data to be processed, so as to classify the data based on the labels and obtain the categories of the data. The categories of the data to be processed include at least one of customer opinions, courier opinions, and customer service opinions.

[0083] A training dataset is generated based on the corpus and its labels. The large language model is then fine-tuned and trained based on this training dataset to obtain the labeling model.

[0084] Thus, in this embodiment, by fine-tuning the large language model based on the acquired corpus and its labels, a labeling model can be obtained. This labeling model can accurately determine the labels of interactive data in logistics services, thereby enabling rapid and accurate identification of the category of the data to be processed based on these labels.

[0085] For example, based on the above data classification and model training methods, the process of automatic labeling and classification can be as follows: Figure 3 As shown, the process includes the following steps:

[0086] Step 1: Acquire a large amount of logistics knowledge, including logistics (non-dialogue) corpora and various logistics dialogue corpora. Based on the acquired logistics knowledge, train a large logistics language model using fine-tuning. This large logistics language model can employ large language modeling techniques such as GPT and Llama frameworks, which have strong semantic understanding capabilities. Adding a large amount of logistics knowledge to the large language model yields a new large logistics language model with even stronger understanding capabilities in the logistics domain.

[0087] Step 2: Using a large corpus of language pairs, train an automatic labeling model based on the logistics big language model to obtain the labeling model.

[0088] The training method is fine-tuning training.

[0089] The corpus consists of corpora and their corresponding tags. The corpus consists of feedback from customers, couriers, customer service, etc. The tags can be a single sentence that highly summarizes the corpus and is used for classification.

[0090] Furthermore, the training is conducted using either single-pair or one-to-many corpus pairs. A single-pair corpus pair means that one piece of corpus corresponds to one label. For example, a courier's feedback, "Why haven't the company uniforms arrived yet?", would be labeled "Unclear delivery time for courier uniforms". A one-to-many corpus pair means that one piece of corpus corresponds to multiple labels. For example, a courier's feedback, "Why haven't the company uniforms arrived yet? And the last uniforms were of very poor quality!", would be labeled with "Unclear delivery time for courier uniforms" and "Courier reports poor uniform quality".

[0091] The tags in the above corpus pairs can be stored in a tag library, including, for example, a customer requesting a second delivery of the package, or a courier reporting that the package is lost. The corresponding corpus is a customer calling to say, "I haven't received the package, please deliver it again tomorrow morning," and a courier reporting, "The package can't be found."

[0092] Of course, depending on the specific needs, a labeling model can also be obtained by training a general large language model.

[0093] Step 3: Automatically tag the current corpus, supporting tagging of existing tags in the tag library, as well as tagging of any new logistics-related tags.

[0094] Specifically, a tag library is used for tagging. If the tagging is successful, the tagging result is obtained. If the tagging fails, a tagging model is used to mine new tags to obtain the tagging result, which is the tag of the corpus.

[0095] This process utilizes a tag library and a logistics language model to determine whether a tag in the library describes the content of the current corpus. For example, if a tag in the library is "customer dissatisfied with delivery time," the logistics language model checks if the corpus contains the content "user dissatisfied with delivery time." If it does, the current corpus is determined to belong to the tag "user dissatisfied with delivery time," and the corpus is tagged under that tag, thus identifying the tag as "user dissatisfied with delivery time." If it does not, the current corpus is determined not to belong to the tag "user dissatisfied with delivery time," and new tags are extracted from the current corpus based on the tagging model.

[0096] Step 4: After any new logistics label is applied, place the new label in the label library.

[0097] Step 5: Push the tagging results to the analysis platform to automatically analyze the main issues, namely the classification of tags, and then determine the category of the current corpus.

[0098] The analysis platform can use a multimodal model for label analysis. The labeling results are input into the multimodal model of the analysis platform for direct analysis. The multimodal model outputs the analysis results, which are used to inform the label classification of the corpus.

[0099] Step 6: Display the labels and categories of all corpora, as well as the content of the corpora under each category.

[0100] The display format is not limited to text, images, and / or tables.

[0101] Taking a tag library containing, for example, 6,000 tags, the analysis platform's multimodal model analyzes which of these 6,000 tags are related to complaints and informs the relevant parties. This can be done by listing tags under each tag category, and vice versa, in a tree diagram / table format. Tag categories include, for example, complaints about damaged goods or complaints about timeliness. This allows management to understand the content of customer complaints and feedback from couriers and customer service, enabling timely resolution of issues. For instance, if the number of tags related to timeliness exceeds a preset limit, measures such as increasing the number of delivery personnel can be taken to reduce the timeliness complaint rate and improve logistics service levels.

[0102] Exemplary device

[0103] like Figure 4 As shown in the figure, this application embodiment also provides a data classification device, including a marking module 401 and a classification module 402.

[0104] in,

[0105] The tagging module 401 is used to tag the data to be processed through the logistics big language model to obtain the tags of the data to be processed. The logistics big language model is a big language model applied to the logistics field, and the data to be processed is interactive data in logistics services.

[0106] The classification module 402 is used to classify the data to be processed based on the tags of the data to be processed to obtain the category of the data to be processed. The category of the data to be processed includes at least one of customer opinions, courier opinions, and customer service opinions.

[0107] The data classification device provided in this embodiment belongs to the same application concept as the data classification method provided in the above embodiments of this application. It can execute the method provided in any of the above embodiments of this application and has the corresponding functional modules and beneficial effects of the method. Technical details not described in detail in this embodiment can be found in the specific processing content of the data classification method provided in the above embodiments of this application, and will not be repeated here.

[0108] The functions implemented by the marking module 401 and the classification module 402 can be implemented by the same or different processors calling software, and this application embodiment does not limit this.

[0109] like Figure 5 As shown in the figure, this application embodiment also provides a model training device, including an acquisition module 501 and a training module 502.

[0110] Among them, the acquisition module 501 is used to acquire logistics knowledge;

[0111] Training module 502 is used to fine-tune the large language model based on the logistics knowledge to obtain a logistics large language model. The logistics large language model is a large language model applied to the logistics field, used to label the data to be processed to obtain labels for the data to be processed, so as to classify the data to be processed based on the labels and obtain the category of the data to be processed. The data to be processed is interactive data in logistics services, and the category includes at least one of customer opinions, courier opinions, and customer service opinions.

[0112] The model training apparatus provided in this embodiment belongs to the same concept as the model training method provided in the above embodiments of this application. It can execute the model training method provided in any of the above embodiments of this application and has the corresponding functional modules and beneficial effects for executing the model training method. Technical details not described in detail in this embodiment can be found in the specific processing content of the model training method provided in the above embodiments of this application, and will not be repeated here.

[0113] The functions implemented by the acquisition module and the training module can be implemented by the same or different processors calling software, and this application embodiment does not limit this.

[0114] This application also provides a model training apparatus, including an acquisition module and a training module.

[0115] The acquisition module is used to acquire the corpus and its tags;

[0116] The training module is used to fine-tune the large language model based on the corpus and its labels to obtain a labeled model. The large language model can be a general model or the aforementioned logistics large language model. The labeled model is used to label the data to be processed, so that the data can be classified based on its labels to obtain the categories of the data to be processed. The categories of the data to be processed include at least one of customer opinions, courier opinions, and customer service opinions.

[0117] The model training apparatus provided in this embodiment belongs to the same concept as the model training method provided in the above embodiments of this application. It can execute the model training method provided in any of the above embodiments of this application and has the corresponding functional modules and beneficial effects for executing the model training method. Technical details not described in detail in this embodiment can be found in the specific processing content of the model training method provided in the above embodiments of this application, and will not be repeated here.

[0118] The functions implemented by the acquisition module and the training module can be implemented by the same or different processors calling software, and this application embodiment does not limit this.

[0119] Exemplary electronic devices

[0120] Another embodiment of this application also provides an electronic device, see [link to relevant documentation] Figure 6 As shown, the electronic device includes a memory 600 and a processor 610.

[0121] The memory 600 is connected to the processor 610 and is used to store programs;

[0122] The processor 610 is used to implement the data classification or model training method disclosed in any of the above embodiments by running the program stored in the memory 600.

[0123] Specifically, the electronic device may also include: a bus, a communication interface 620, an input device 630, and an output device 640.

[0124] The processor 610, memory 600, communication interface 620, input device 630, and output device 640 are interconnected via a bus. Among them:

[0125] A bus can include a pathway for transmitting information between various components of a computer system.

[0126] The processor 610 can be a general-purpose processor, such as a general-purpose central processing unit (CPU), a microprocessor, etc., or an application-specific integrated circuit (ASIC), or one or more integrated circuits used to control the execution of the program of the present application. It can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0127] The processor 610 may include a main processor, as well as a baseband chip, modem, etc.

[0128] The memory 600 stores a program for executing the technical solution of this application, and may also store an operating system and other critical business functions. Specifically, the program may include program code, which includes computer operation instructions. More specifically, the memory 600 may include read-only memory (ROM), other types of static storage devices capable of storing static information and instructions, random access memory (RAM), other types of dynamic storage devices capable of storing information and instructions, disk storage, flash memory, etc.

[0129] Input device 630 may include a device for receiving user input data and information, such as a keyboard, mouse, camera, scanner, light pen, voice input device, touch screen, pedometer, or gravity sensor.

[0130] Output device 640 may include devices that allow information to be output to a user, such as a display screen, printer, speaker, etc.

[0131] The communication interface 620 may include a device that uses any transceiver to communicate with other devices or communication networks, such as Ethernet, Radio Access Network (RAN), Wireless Local Area Network (WLAN), etc.

[0132] The processor 610 executes the program stored in the memory 600 and calls other devices, which can be used to implement any of the steps of the data classification or model training method provided in the above embodiments of this application.

[0133] Those skilled in the art will understand that Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the electronic device to which the present application is applied. The specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0134] This application also proposes a chip including a processor and a data interface. The processor reads and runs a program stored in a memory through the data interface to execute the data classification or model training methods described in any of the above embodiments. For details of the processing and its beneficial effects, please refer to the embodiments of the above data classification or model training methods.

[0135] In addition to the methods and apparatus described above, embodiments of this application propose a computer program product comprising computer program instructions that, when executed by a processor, cause the processor to perform the steps in the data classification or model training methods according to various embodiments of this application as described in the "Exemplary Methods" section of this specification.

[0136] The computer program product can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of this application. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0137] Furthermore, embodiments of this application also propose a storage medium storing a computer program, which is executed by a processor in the steps of the data classification or model training methods according to various embodiments of this application described in the "Exemplary Methods" section above.

[0138] The basic principles of the present invention have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in the present invention are merely examples and not limitations, and should not be considered as essential features of each embodiment of the present invention. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the present invention to the necessity of employing the aforementioned specific details.

[0139] The block diagrams of devices, apparatuses, devices, and systems involved in this invention are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0140] It should also be noted that in the apparatus, device, and method of the present invention, the components or steps can be disassembled and / or recombined. These disassemblies and / or recombinations should be considered as equivalent solutions of the present invention.

[0141] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the invention. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of the invention. Therefore, the invention is not intended to be limited to the aspects shown herein, but rather to be carried out within the widest scope consistent with the principles and novel features disclosed herein.

[0142] It should be understood that the qualifying terms "first", "second", "third", "fourth", "fifth" and "sixth" used in the description of the embodiments of the present invention are only used to more clearly illustrate the technical solutions and are not intended to limit the scope of protection of the present invention.

[0143] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of the invention to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations therein.

Claims

1. A data classification method, characterized in that, The method includes: The data to be processed is tagged using a logistics big language model to obtain the labels of the data to be processed. The logistics big language model is a big language model applied to the logistics field, and the data to be processed is interactive data in logistics services. Based on the tags of the data to be processed, the data to be processed is classified to obtain the category of the data to be processed. The category includes at least one of customer opinions, courier opinions, and customer service opinions.

2. The data classification method according to claim 1, characterized in that, The process of tagging the data to be processed using a logistics big data language model to obtain the tags for the data to be processed includes: Using the logistics big language model, the tags in the tag database are matched with the data to be processed to obtain matching results. The matching results include the tags in the tag database that match the data to be processed. Based on the matching results, the label of the data to be processed is determined.

3. The data classification method according to claim 2, characterized in that, Determining the label of the data to be processed based on the matching result includes: If the matching result is not empty, then the tags in the tag database that match the content of the data to be processed are determined as the tags of the data to be processed; If the matching result is empty, the data to be processed is tagged using a tagging model to obtain the label of the data to be processed.

4. The data classification method according to claim 3, characterized in that, After the data to be processed is labeled using a labeling model to obtain the labels for the data to be processed, the method further includes: The tags of the data to be processed are stored in the tag database.

5. The data classification method according to claim 1, characterized in that, The process of classifying the data to be processed based on its labels to obtain the categories of the data to be processed includes: Determine the tag category corresponding to the tag of the data to be processed; The tags corresponding to the tags of the data to be processed are classified and determined as the categories of the data to be processed.

6. The data classification method according to claim 5, characterized in that, The method further includes: The system statistically analyzes and displays the tag categories corresponding to all tags in the tag database.

7. A model training method, characterized in that, The method includes: Acquire logistics knowledge; Based on the logistics knowledge, the large language model is fine-tuned to obtain a logistics large language model. The logistics large language model is a large language model applied to the logistics field. It is used to label the data to be processed to obtain the labels of the data to be processed, so as to classify the data to be processed based on the labels of the data to be processed and obtain the category of the data to be processed. The data to be processed is interactive data in logistics services, and the category includes at least one of customer opinions, courier opinions, and customer service opinions.

8. An electronic device, characterized in that, Including memory and processor; The memory is connected to the processor and is used to store programs; The processor is used to implement the data classification method as described in any one of claims 1 to 6 or the model training method as described in claim 7 by running the program in the memory.

9. A storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the data classification method as described in any one of claims 1 to 6 or the model training method as described in claim 7.

10. A computer program product, characterized in that, The computer program product includes computer program instructions that, when executed by a processor, cause the processor to perform the data classification method as described in any one of claims 1-6 or the model training method as described in claim 7.

Images