A work order processing method based on community scenarios and related components

Abstract

The present invention discloses a work order processing method based on a community scenario and related components. The method constructs a sequence annotation model based on the work order corpus, and constructs a classification model based on the work order classification corpus and the work order corpus; when a target work order is received, the sequence annotation model is used to extract features and annotate the target work order to obtain multiple entity information; then the classification model is used to classify the target work order to obtain a classification label; a work order task is generated based on the entity information and classification label of the target work order and dispatched to the work order handler. The method achieves the advantage of accurately classifying and dispatching work orders based on community scenarios.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a work order processing method based on a community scenario and related components. Background Art

[0002] Community service agencies in community scenarios usually receive a large number of customer-oriented work orders. Traditional community work orders are sent to community service agencies via telephone / text messages / APP communications, etc. The agency’s customer service specialists manually determine the content of the information and then distribute it to the work order handler to handle the matter and further verify the work order.

[0003] However, in large communities and high-frequency work order scenarios, the traditional community work order system has low operating efficiency and is difficult to meet the needs of high-quality community operation scenarios. Ordinary work order classification systems are difficult to meet the needs of diverse and personalized community service types. Summary of the invention

[0004] The purpose of the present invention is to provide a work order processing method and related components based on community scenarios, aiming to solve the problem of low efficiency and accuracy of work order allocation and intelligent dispatch in community scenarios in the prior art.

[0005] In a first aspect, an embodiment of the present invention provides a work order processing method based on a community scenario, which includes:

[0006] Performing text preprocessing on the work order corpus to obtain a first text sequence;

[0007] The first text sequence is vectorized by using a language representation model, and then the vector of the first text sequence is learned by using a named entity recognition algorithm to obtain a plurality of entity information, thereby constructing a sequence labeling model;

[0008] Construct a work order classification corpus according to the service type, perform text sequence representation and vectorization on the work order classification corpus and the work order corpus, output a vector of the second text sequence, and then input it into the LSTM neural network model, output deep semantic features, and then connect it to the softmax activation function for multi-classification training to obtain classification labels, thereby constructing a classification model;

[0009] When a target work order is received, the target work order is annotated using the sequence annotation model to obtain a plurality of entity information; and the target work order is classified using the classification model to obtain a classification label;

[0010] A work order task is generated according to the entity information and classification label of the target work order and is dispatched to the work order handler.

[0011] In a second aspect, an embodiment of the present invention provides a work order processing device based on a community scenario, which includes:

[0012] A preprocessing unit, used for performing text preprocessing on the work order corpus to obtain a first text sequence;

[0013] A first construction unit performs vectorization processing on the first text sequence through a language representation model, and then uses a named entity recognition algorithm to learn the vector of the first text sequence to obtain multiple entity information, thereby constructing a sequence labeling model;

[0014] The second construction unit is used to construct a work order classification corpus according to the service type, perform text sequence representation and vectorization processing on the work order classification corpus and the work order corpus, output a vector of the second text sequence, and then input it into the LSTM neural network model, output deep semantic features, and then connect it to the softmax activation function for multi-classification training to obtain classification labels, thereby constructing a classification model;

[0015] A classification unit, configured to, when receiving a target work order, label the target work order using the sequence labeling model to obtain a plurality of entity information; and classify the target work order using the classification model to obtain a classification label;

[0016] The dispatching unit is used to generate a work order task according to the entity information and classification label of the target work order and dispatch it to the work order handler.

[0017] In the third aspect, an embodiment of the present invention further provides a computer device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the community scenario-based work order processing method described in the first aspect is implemented.

[0018] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor executes the community scenario-based work order processing method described in the first aspect above.

[0019] The present invention discloses a work order processing method based on a community scenario and related components. The method constructs a sequence annotation model based on the work order corpus, and constructs a classification model based on the work order classification corpus and the work order corpus; when a target work order is received, the sequence annotation model is used to extract features and annotate the target work order to obtain multiple entity information; then the classification model is used to classify the target work order to obtain a classification label; a work order task is generated based on the entity information and classification label of the target work order and dispatched to the work order handler. The embodiment of the present invention achieves the advantage of accurately classifying and dispatching work orders based on community scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying any creative work.

[0021] Figure 1 A flowchart of a community scenario-based work order processing method provided by an embodiment of the present invention;

[0022] Figure 2 A schematic diagram of a sub-process of a work order processing method based on a community scenario provided in an embodiment of the present invention;

[0023] Figure 3 A schematic diagram of another sub-process of the work order processing method based on the community scenario provided by an embodiment of the present invention;

[0024] Figure 4 A schematic diagram of another sub-process of the work order processing method based on the community scenario provided by an embodiment of the present invention;

[0025] Figure 5 A schematic diagram of another sub-process of the work order processing method based on the community scenario provided by an embodiment of the present invention;

[0026] Figure 6 A schematic diagram of another sub-process of the work order processing method based on the community scenario provided by an embodiment of the present invention;

[0027] Figure 7 A schematic diagram of a named entity recognition corpus provided by an embodiment of the present invention;

[0028] Figure 8 A schematic diagram of classification model training provided by an embodiment of the present invention;

[0029] Fig. 9 A schematic diagram of a work order classification corpus provided by an embodiment of the present invention;

[0030] Fig.10 A schematic block diagram of a work order processing device based on a community scenario provided by an embodiment of the present invention;

[0031] Fig.11 A schematic block diagram of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0032] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0033] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0034] It should also be understood that the terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0035] It should be further understood that the term "and / or" used in the present description and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0036] See also Figure 1 , Figure 1 A flowchart of a community scenario-based work order processing method provided by an embodiment of the present invention;

[0037] like Figure 1 As shown, the method includes steps S101 to S104.

[0038] S101. Perform text preprocessing on the work order corpus to obtain a first text sequence.

[0039] In this embodiment, the work order corpus is a large amount of work order text data. The content of the work order text is first read, and then the work order text is segmented by tools such as NLPIR Chinese word segmentation software and THULAC Chinese word segmentation software, and abnormal characters are removed to obtain a first text sequence.

[0040] In one embodiment, if Figure 2 As shown, step S101 includes:

[0041] S201, pre-build named entity recognition corpus based on community scenarios;

[0042] S202, merging the work order corpus with the named entity recognition corpus to obtain a named entity sequence corpus set;

[0043] S203: Remove abnormal characters from the named entity sequence corpus to obtain a first text sequence.

[0044] In this embodiment, in the process of constructing a named entity recognition corpus based on community scenarios, a large number of corpus texts based on different community scenarios are integrated, named entities are extracted from the corpus texts and a named entity recognition corpus is constructed, such as named entities such as owners, tenants, parking lots, etc. that can be extracted from the corpus texts based on the community.

[0045] Combination Figure 7 As shown, the work order text data in the work order corpus contains a large number of named entities such as personal names PER, roles ROLE, addresses LOC, and organizations ORG. The named entity recognition corpus based on the community scenario contains specific personal names PER (such as Zhang San, Li Si), specific roles ROLE (such as housekeeper, owner), specific addresses LOC (such as underground parking lot), specific organizations ORG (such as service center), etc.; the work order corpus is merged with the named entity recognition corpus to obtain a named entity sequence corpus set; then, abnormal characters (such as spaces, symbols, etc.) in the named entity sequence corpus are removed, and the word frequency in the repeated named entities is calculated, and words with too few appearances are excluded, so as to reduce the text dimension and obtain the first text sequence.

[0046] S102: vectorize the first text sequence using a language representation model, and then use a named entity recognition algorithm to learn the vector of the first text sequence to obtain multiple entity information, thereby constructing a sequence labeling model.

[0047] In this embodiment, for the named entity recognition algorithm, its input is a vector, so it is necessary to first vectorize the first text sequence through the language representation model in order to recognize the first text sequence, and input the vector of the first text sequence into the named entity recognition algorithm for learning to obtain multiple entity information, thereby constructing a sequence labeling model.

[0048] In one embodiment, if Figure 3 As shown, step S102 includes:

[0049] S301, performing vectorization processing on the first text sequence through a BERT model to obtain a vector of the first text sequence;

[0050] S302: Input the vector of the first text sequence into BILSTM-CRF for feature extraction and labeling to obtain multiple entity information;

[0051] S303, training through the loss function to build a sequence labeling model.

[0052] In this embodiment, the BERT model (Bidirectional Encoder Representations from Transformers, BERT for short) is one of the language representation models, and the Word2Vec model or the GPT-2 model can also be selected here. The first text sequence can also be vectorized and the vector of the first text sequence can be obtained; the BILSTM-CRF is one of the named entity recognition algorithms, and HMM (Hidden Markov model, HMM for short) or linear-CRF (Linear chain Conditional Random Fields, linear-CRF for short) can also be selected here. The vector of the first text sequence can also be feature extracted and labeled, and then multiple entity information is obtained, and the entity information refers to the name of a person, the name of an organization, the address and time and the corresponding output label in a specific community scenario; finally, the loss function is used for training to construct a sequence labeling model.

[0053] Specifically, take the text A in the named entity sequence corpus: "I am a resident of Room 201, Unit 5. My faucet is broken and I want to hire someone to repair it." as an example; first, text A is preprocessed and a text sequence of text A is obtained, and then the text sequence of text A is vectorized through the BERT model to obtain a text sequence vector with a length of 512, and then the text sequence vector is input into the BILSTM-CRF for feature extraction and annotation to obtain multiple entity information.

[0054] The entity information includes the processing location "Unit 5, 201"; the work order initiator "the resident of Unit 5, 201"; the maintenance material involved "faucet"; and the corresponding output label: "[CLS][CLS] IB-PER\ am O\5B-LOC\ unit I-LOC\ unit I-LOC\2I-LOC\0I-LOC\1I-LOC\ O\ resident B-ROLE\ household I-ROLE, the O\ water B-MAT\ faucet I-MAT\ in my B-LOC\ home I-LOC\ is broken O\, I want O\ to ask O\ someone O\ to come O\ to repair O\ [SEP][SEP]".

[0055] Based on the BILSTM-CRF, feature extraction and annotation are performed on the entire named entity sequence corpus to obtain a large amount of entity information. Finally, the absolute difference range t is set by the loss function. When the absolute value of the difference between the predicted label value and the true label value is less than t, the two are considered to be the same, and the loss function is 0 at this time; when the absolute value of the difference between the predicted label value and the true label value is greater than t, the two are considered to be different, and the loss function is 1 at this time, thereby constructing a sequence annotation model.

[0056] S103. Construct a work order classification corpus according to the service type, perform text sequence representation and vectorization on the work order classification corpus and the work order corpus, output a vector of the second text sequence, and then input it into the LSTM neural network model, output deep semantic features, and then connect to the softmax activation function for multi-classification training to obtain classification labels, thereby constructing a classification model.

[0057] like Figure 8 As shown, in this embodiment, the service type is used as a label to construct a work order classification corpus, the service type is set according to a specific community scenario, the work order classification corpus and the work order corpus are represented by a text sequence through a BERT model to obtain a second text sequence, and then the second text sequence is vectorized to obtain a vector of the second text sequence, and then the vector of the second text sequence is input into an LSTM neural network model to perform feature extraction and output deep semantic features, and after feature extraction, a softmax activation function is connected to perform multi-classification training on the obtained deep semantic features to obtain classification labels, thereby constructing a classification model for classifying target work orders.

[0058] In one embodiment, if Figure 4 As shown, the work order classification corpus is constructed according to the service type, including:

[0059] S401, classify the service types in advance to obtain service types of different levels;

[0060] S402: Setting subordinate relationships for service types at different levels, thereby constructing a work order classification corpus.

[0061] In this embodiment, the service types can be divided into multiple levels to classify the types of work orders more systematically. In this embodiment, the service types are divided into three levels for illustration (eg Fig. 9), where the first-level classification includes "maintenance", "complaint", "praise" and "service", where the service type "maintenance" includes multiple second-level service types, specifically: "house maintenance", "home appliance maintenance", "installation and removal" and "outdoor air conditioning", and the service type "house maintenance" includes multiple third-level service types, specifically: "lock opening and replacement", "pipe unblocking", "waterway maintenance" and "furniture maintenance"; after different service types are graded and subordinate relationships are set, the work order classification corpus can be constructed.

[0062] In one embodiment, if Figure 5 As shown, step S103 includes:

[0063] S501. Constructing work order classification corpus according to service type;

[0064] S502, performing text sequence representation and vectorization processing on the work order classification corpus and the work order corpus through BERT, and outputting a vector of a second text sequence;

[0065] S503, inputting the vector of the second text sequence into a 3-layer LSTM neural network model and outputting deep semantic features, wherein a generalization layer is added between each layer of the neural network in the LSTM neural network model;

[0066] S504, connecting the output deep semantic features to the softmax activation function for multi-classification training to obtain classification labels, thereby building a classification model.

[0067] In this embodiment, the service type is used as a label to construct a work order classification corpus, and the service type is set according to different community scenarios. For example, in a residential community scenario, service types such as house maintenance, water pipeline maintenance, and refrigerator cleaning can be set. These service types based on the community scenario and the names PER, roles ROLE, addresses LOC, and organizations ORG in the work order corpus are represented by a text sequence through a BERT model to obtain a second text sequence, and then the second text sequence is vectorized to obtain a vector of the second text sequence, and then the vector of the second text sequence is input into a 3-layer LSTM neural network model for feature extraction, and a generalization layer is added between each layer of the neural network to improve the performance of the neural network model, and deep semantic features are output. After feature extraction, a softmax activation function is connected to perform multi-classification training on the obtained deep semantic features to obtain classification labels, thereby constructing a classification model for classifying target work orders.

[0068] Specifically, take the above text A: "I am a resident of Room 201, Unit 5. My faucet is broken and I want to hire someone to repair it." Continuing with the example, based on the entity information in text A, the service type is output: "Water Pipe Repair", and then based on a large amount of text, multiple labels are output and a classification model is constructed to classify the target work order.

[0069] In one embodiment, if Figure 6 As shown, the vector of the second text sequence is input into a 3-layer LSTM neural network model and the deep semantic features are output, including:

[0070] S601, inputting the vector of the second text sequence into a 3-layer LSTM neural network model to obtain weights W and biases b of different paths;

[0071] S602: Use the weights W and biases b of different paths to fuse multiple features to obtain deep semantic features.

[0072] In this embodiment, during the classification model training process, the training target of the CNN pathway is the weight W CNN and bias b CNN ; The training target of the LSTM path is the weight W LSTM and bias b LSTM ; The training target of the BERT pathway is W BERT and bias b BERT ; During model training, data is forward propagated along the deep neural network model, and all neural networks fuse multivariate features at output; finally, the parameters in the classification model are trained and obtained:

[0073] Q={W CNN , W LSTM , W BERT , b CNN , b LSTM , b BERT}.

[0074] In one embodiment, step S103 further includes:

[0075] The Adam algorithm is used for back propagation, and the parameters in the classification model are updated for multiple rounds to make the classification accuracy reach the preset threshold.

[0076] In this embodiment, during back propagation, the partial derivative of Q is calculated using the loss function Loss, and the parameters in the classification model are updated so that the classification accuracy reaches a preset threshold. The back propagation method used is the Adam algorithm (adaptive moment estimation, referred to as Adam); the SGD algorithm (Stochastic gradient descent, referred to as SGD) or the Momentum algorithm (gradient descent with momentum, referred to as Momentum, momentum gradient descent) can also be used here.

[0077] S104. When a target work order is received, the target work order is labeled using the sequence labeling model to obtain a plurality of entity information; and the target work order is classified using the classification model to obtain a classification label.

[0078] In this embodiment, when a target work order is received, the text of the target work order is first preprocessed to obtain a target text sequence of the target work order, and then the target text sequence is vectorized through the BERT model to obtain a vector of the target text sequence, and then the vector of the target text sequence is input into the sequence annotation model for sign extraction and annotation to obtain entity information based on the target text sequence; and then the vector of the target text sequence is input into the classification model for work order classification and output of the classification label of the target work order.

[0079] Specifically, taking the target work order "I am the resident of unit 3, 312. My door lock is broken and I want to ask someone to repair it." as an example, after the vector of the target text sequence is input into the sequence labeling model for sign extraction and labeling, entity information is obtained, and the entity information includes the processing location "unit 3, 312"; the work order initiator "the resident of unit 3, 312"; the involved repair material "door lock"; and the corresponding output label: "[CLS][CLS] I B-PER\ am O\3B-LOC\ unit I-LOC\ unit I-LOC\ 3I-LOC\ 1I-LOC\ 2I-LOC\ O\ resident B-ROLE\ household I-ROLE, the O\ door B-MAT\ lock I-MAT\ of my B-LOC\ home I-LOC\ is broken O\, I\ want O\ to ask O\ someone O\ to O\ repair O\ [SEP][SEP]"

[0080] Then, the vector of the target text sequence is input into the classification model to classify the work order and output the classification label of the target work order: opening and changing locks; thus, the entity information and classification label of the target work order can be obtained.

[0081] S105: Generate a work order task according to the entity information and classification label of the target work order and distribute it to the work order handler.

[0082] In this embodiment, the work order task is dispatched to the recommended work order handler through the work order dispatching module, and information is received from the work order handler on whether the work order task can be handled. If so, the detailed task of the work order is confirmed, and information such as the contact information of the work order initiator and the task completion deadline are sent to the recommended work order handler, so that the recommended work order handler can execute and complete the work order task.

[0083] Furthermore, after the work order processor has processed the target work order, the type of the work order is confirmed, and the work order is cancelled after the confirmation is correct.

[0084] The embodiment of the present invention also provides a work order processing device based on a community scenario, and the work order processing device based on a community scenario is used to execute any embodiment of the work order processing method based on a community scenario. Fig.10 , Fig.10 It is a schematic block diagram of a work order processing device based on a community scenario provided in an embodiment of the present invention.

[0085] like Fig.10 As shown, the work order processing device 1000 based on the community scenario includes: a preprocessing unit 1001, a first construction unit 1002, a second construction unit 1003, a classification unit 1004 and a dispatching unit 1005.

[0086] A preprocessing unit 1001 is used to perform text preprocessing on the work order corpus to obtain a first text sequence;

[0087] The first construction unit 1002 performs vectorization processing on the first text sequence through a language representation model, and then uses a named entity recognition algorithm to learn the vector of the first text sequence to obtain multiple entity information, thereby constructing a sequence labeling model;

[0088] The second construction unit 1003 is used to construct a work order classification corpus according to the service type, perform text sequence representation and vectorization processing on the work order classification corpus and the work order corpus, output a vector of the second text sequence, and then input it into the LSTM neural network model, output deep semantic features, and then connect it to the softmax activation function for multi-classification training to obtain classification labels, thereby constructing a classification model;

[0089] The classification unit 1004 is used for, when receiving a target work order, labeling the target work order using the sequence labeling model to obtain a plurality of entity information; and classifying the target work order using the classification model to obtain a classification label;

[0090] 1005 dispatching unit, used to generate a work order task according to the entity information and classification label of the target work order and dispatch it to the work order handler.

[0091] The device achieves the advantage of accurately classifying and dispatching work orders based on community scenarios through sequence labeling models and classification models.

[0092] Since the embodiments of the apparatus part correspond to the embodiments of the method part, please refer to the description of the embodiments of the method part for the embodiments of the apparatus part, which will not be repeated here.

[0093] The above-mentioned work order processing device based on community scenario can be implemented in the form of a computer program, and the computer program can be used in Fig.11 Runs on the computer device shown.

[0094] See also Fig.11 , Fig.11 11 is a schematic block diagram of a computer device provided by an embodiment of the present invention. The computer device 1100 is a server, which may be an independent server or a server cluster composed of multiple servers.

[0095] See also Fig.11 The computer device 1100 includes a processor 1102 , a memory, and a network interface 1105 connected via a system bus 1101 , wherein the memory may include a non-volatile storage medium 1103 and an internal memory 1104 .

[0096] The non-volatile storage medium 1103 can store an operating system 11031 and a computer program 11032. When the computer program 11032 is executed, the processor 1102 can execute a work order processing method based on a community scenario.

[0097] The processor 1102 is used to provide computing and control capabilities to support the operation of the entire computer device 1100 .

[0098] The internal memory 1104 provides an environment for the operation of the computer program 11032 in the non-volatile storage medium 1103. When the computer program 11032 is executed by the processor 1102, the processor 1102 can execute a work order processing method based on a community scenario.

[0099] The network interface 1105 is used for network communication, such as providing data information transmission, etc. Those skilled in the art will understand that Fig.11The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device 1100 to which the solution of the present invention is applied. The specific computer device 1100 may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0100] Those skilled in the art will understand that Fig.11 The embodiments of the computer device shown in the figure do not constitute a limitation on the specific composition of the computer device. In other embodiments, the computer device may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently. For example, in some embodiments, the computer device may only include a memory and a processor. In such embodiments, the structure and function of the memory and the processor are the same as those of the embodiment of the present invention. Fig.11 The embodiments shown are consistent and will not be described again here.

[0101] It should be understood that in the embodiment of the present invention, the processor 1102 may be a central processing unit (CPU), and the processor 1102 may also be a graphics processing unit (GPU), a tensor processing unit (TPU), other general-purpose processors, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0102] In another embodiment of the present invention, a computer-readable storage medium is provided. The computer-readable storage medium may be a non-volatile computer-readable storage medium. The computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, the work order processing method based on the community scenario of an embodiment of the present invention is implemented.

[0103] The storage medium is a physical, non-transient storage medium, for example, it can be a USB flash drive, a mobile hard disk, a read-only memory (ROM), a magnetic disk or an optical disk, etc., which can store program codes.

[0104] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described equipment, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0105] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A work order processing method based on community scenarios, characterized in that: include: Performing text preprocessing on the work order corpus to obtain a first text sequence; The first text sequence is vectorized by using a language representation model, and then the vector of the first text sequence is learned by using a named entity recognition algorithm to obtain a plurality of entity information, thereby constructing a sequence labeling model; Construct a work order classification corpus according to the service type, perform text sequence representation and vectorization on the work order classification corpus and the work order corpus, output a vector of the second text sequence, and then input it into the LSTM neural network model, output deep semantic features, and then connect it to the softmax activation function for multi-classification training to obtain classification labels, thereby constructing a classification model; When a target work order is received, the target work order is annotated using the sequence annotation model to obtain a plurality of entity information; and the target work order is classified using the classification model to obtain a classification label; Generate a work order task according to the entity information and classification label of the target work order and dispatch it to the work order handler; Among them, the work order classification corpus is constructed according to the service type, the work order classification corpus and the work order corpus are represented and vectorized as text sequences, the vector of the second text sequence is output, and then it is input into the LSTM neural network model, the deep semantic features are output, and then it is connected to the softmax activation function for multi-classification training to obtain classification labels, thereby constructing a classification model, including: constructing the work order classification corpus according to the service type; representing and vectorizing the work order classification corpus and the work order corpus as text sequences through BERT, and outputting the vector of the second text sequence; inputting the vector of the second text sequence into a 3-layer LSTM neural network model and outputting deep semantic features, wherein a generalization layer is added between each layer of the neural network in the LSTM neural network model; connecting the output deep semantic features to the softmax activation function for multi-classification training to obtain classification labels, thereby constructing a classification model.

2. The work order processing method based on community scenario according to claim 1 is characterized in that: The step of performing text preprocessing on the work order corpus to obtain a first text sequence includes: Pre-build named entity recognition corpus based on community scenarios; Merging the work order corpus with the named entity recognition corpus to obtain a named entity sequence corpus set; Abnormal characters in the named entity sequence corpus are removed to obtain a first text sequence.

3. The work order processing method based on community scenario according to claim 1 is characterized in that: The vectorization processing of the first text sequence is performed by using a language representation model, and then the vector of the first text sequence is learned by using a named entity recognition algorithm to obtain a plurality of entity information, thereby constructing a sequence labeling model, including: Vectorize the first text sequence using the BERT model to obtain a vector of the first text sequence; Input the vector of the first text sequence into BILSTM-CRF for feature extraction and labeling to obtain multiple entity information; The loss function is used for training to build a sequence labeling model.

4. The work order processing method based on community scenario according to claim 1 is characterized in that: The construction of the work order classification corpus according to the service type includes: Classify the service types in advance to obtain different levels of service types; Set up subordinate relationships for different levels of service types to build a work order classification corpus.

5. The work order processing method based on community scenario according to claim 1 is characterized in that: The method constructs a work order classification corpus according to the service type, performs text sequence representation and vectorization processing on the work order classification corpus and the work order corpus, outputs a vector of a second text sequence, and then inputs it into an LSTM neural network model, outputs deep semantic features, and then connects to a softmax activation function for multi-classification training to obtain classification labels, thereby constructing a classification model, and also includes: The Adam algorithm is used for back propagation, and the parameters in the classification model are updated for multiple rounds to make the classification accuracy reach the preset threshold.

6. The work order processing method based on community scenario according to claim 1 is characterized in that: The step of inputting the vector of the second text sequence into a 3-layer LSTM neural network model and outputting deep semantic features includes: Input the vector of the second text sequence into a 3-layer LSTM neural network model to obtain weights W and biases b of different paths; The weights W and bias b of different paths are used to fuse multiple features to obtain deep semantic features.

7. A work order processing device based on community scenarios, characterized in that: include: A preprocessing unit, used for performing text preprocessing on the work order corpus to obtain a first text sequence; A first construction unit performs vectorization processing on the first text sequence through a language representation model, and then uses a named entity recognition algorithm to learn the vector of the first text sequence to obtain multiple entity information, thereby constructing a sequence labeling model; The second construction unit is used to construct a work order classification corpus according to the service type, perform text sequence representation and vectorization processing on the work order classification corpus and the work order corpus, output a vector of the second text sequence, and then input it into the LSTM neural network model, output deep semantic features, and then connect it to the softmax activation function for multi-classification training to obtain classification labels, thereby constructing a classification model; A classification unit, configured to, when receiving a target work order, label the target work order using the sequence labeling model to obtain a plurality of entity information; and classify the target work order using the classification model to obtain a classification label; A dispatching unit, configured to generate a work order task according to the entity information and classification label of the target work order and dispatch the task to a work order handler; Among them, the second construction unit is used to construct a work order classification corpus according to the service type; the work order classification corpus and the work order corpus are represented and vectorized by BERT, and the vector of the second text sequence is output; the vector of the second text sequence is input into a 3-layer LSTM neural network model and deep semantic features are output, wherein a generalization layer is added between each layer of the neural network in the LSTM neural network model; the output deep semantic features are connected to the softmax activation function for multi-classification training to obtain classification labels, thereby constructing a classification model.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, it implements the work order processing method based on the community scenario as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed by a processor, enables the processor to execute the community scenario-based work order processing method as described in any one of claims 1 to 6.

Images