Community demand intelligent matching service system based on natural language processing

Abstract

The invention relates to the technical field of artificial intelligence and community demand matching, and discloses a community demand intelligent matching service system based on natural language processing, which comprises a text vector generation module, a multi-source vector fusion module, a knowledge graph analysis module, a candidate service optimization module and a service matching push module. In the prior art, a community service recommendation mode which depends on keyword or fixed template matching, and especially under the condition that resident expression is blurred or context changes frequently, accurate identification and intelligent recommendation of service intentions are difficult to realize. According to the method, the natural language pre-training model is introduced, and the community service knowledge graph and the multilayer semantic review sorting mechanism are constructed, so that semantic understanding and personalized service pushing of resident service requirements are realized, the problems of large recommendation deviation, low service coverage and poor response efficiency are avoided, and the user experience is improved. And the intelligence and the response accuracy of the community service system are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of artificial intelligence and community demand matching, and in particular relates to a community demand intelligent matching service system based on natural language processing. Background Art

[0002] Currently, community governance and service response systems rely heavily on manual screening, fixed-template questionnaires, and keyword matching to identify resident needs and recommend services. However, these methods are significantly inadequate for addressing residents' unstructured and ambiguous expressions. For example, residents often express their needs through voice, images, or text messages via community apps, service hotlines, or during visits. These expressions often contain semantic ambiguity, jumpy expressions, or logical incoherence. Traditional service dispatch mechanisms based on keyword rules or process node triggers struggle to accurately identify their true service intent, resulting in inaccurate service recommendations, delayed responses, and even mismatched services, severely impacting the intelligent level of community services and resident satisfaction. Existing technologies cannot fully meet the intelligent identification and precise recommendation requirements for community service applications involving "low-expressive residents," "complex semantic expression scenarios," and "multi-channel heterogeneous data fusion." In particular, in situations with strong emotional expressions, ambiguous context, and interdependent service hierarchies, service intent is difficult to accurately interpret, resulting in low adaptability and coverage of system recommendations. Therefore, there is an urgent need for an intelligent community service matching method that can achieve service intention semantic modeling, service node structure matching, and service priority sorting optimization even when residents' expressions are complex and service structures are densely correlated, so as to improve the community service system's ability to understand residents' actual needs, response accuracy, and the ability to coordinate and link multiple services. Summary of the Invention

[0003] In response to the above-mentioned technical deficiencies, the purpose of the present invention is to propose a community demand intelligent matching service system and method based on natural language processing, aiming to solve the technical problem that the existing community service recommendation method relies heavily on keywords or fixed template matching, especially when residents' expressions are unclear or the context changes frequently, it is difficult to accurately identify service intentions and make intelligent recommendations.

[0004] In order to solve the above technical problems, the present invention adopts the following technical solutions: The present invention provides a community demand intelligent matching service system based on natural language processing,

[0005] The community demand intelligent matching service system based on natural language processing includes:

[0006] The text vector generation module is used to obtain the original expression corpus of the current service window from the community server, perform OCR recognition and standardization on the original expression corpus, and generate standard expression sentences. A natural language pre-training model is pre-set and trained, and the standard expression sentences are input into the natural language pre-training model to output the text expression vector.

[0007] The multi-source vector fusion module is used to obtain the current resident portrait feature vector and the current service environment feature vector from the community service end, and perform weighted fusion processing on them in combination with the text expression vector to obtain the fused demand expression vector;

[0008] The knowledge graph analysis module is used to preset the community service knowledge graph, perform semantic matching on the community service knowledge graph based on the fusion demand expression vector, and generate a preliminary set of candidate services;

[0009] The candidate service optimization module is used to set up a three-layer natural language review model, sort the preliminary candidate service set based on the three-layer natural language review model and combine it with a multi-objective ranking optimization method to generate an optimized candidate service set;

[0010] The service matching push module is used to generate service recommendation cards based on the optimized candidate service set and push the service recommendation cards to the resident end.

[0011] Preferably, in the text vector generation module, the original expression corpus specifically includes hotline voice transcription text, community app messages, WeChat text and picture combination data, interview forms and visit records, mini-program search bar input, open-ended answers recognized by the AI outbound call system, handwritten demand forms, medical records and paper application forms.

[0012] Preferably, in the text vector generation module, the structure of the natural language pre-training model specifically includes:

[0013] The input encoding layer is used to encode standard expression sentences into vector inputs that can be processed by the natural language pre-training model;

[0014] Multi-layer Transformer context modeling layer, including a multi-head self-attention mechanism network, a feedforward neural network, and a residual connection and layer normalization network;

[0015] The fine-tuning mechanism processing layer is used to obtain community-domain annotated corpora and supports lightweight fine-tuning of the semantic vectors output by the multi-layer Transformer context modeling layer using community-domain annotated corpora;

[0016] The output layer is used to extract the semantic vector of the entire sentence from the semantic vector output by the fine-tuning mechanism processing layer, and output the text expression vector.

[0017] Preferably, in the multi-source vector fusion module, the step of obtaining the current resident portrait feature vector and the current service environment feature vector from the community server, and performing weighted fusion processing in combination with the text expression vector to obtain the fused demand expression vector specifically includes:

[0018] Obtain the current resident portrait feature vector and the current service environment feature vector from the community server; wherein, the current resident portrait feature vector is used to represent the current resident portrait; the current service environment feature vector is used to represent the environmental features of the current service, including time, channel, weather, and emergency status;

[0019] First, perform preliminary weighted fusion on the text expression vector, the current resident portrait feature vector, and the current service environment feature vector through the three-vector weighted method to obtain a preliminary demand expression vector;

[0020] Then, use the deep neural network Transformer to further perform non-linear expression expansion processing on the preliminary demand expression vector to obtain the fused demand expression vector.

[0021] Preferably, in the knowledge graph analysis module, the step of presetting the community service knowledge graph and generating a preliminary candidate service set through semantic matching of the fused demand expression vector by the community service knowledge graph specifically includes:

[0022] Preset the community service knowledge graph G, G = (V, E); wherein, V is the set of service nodes, used to represent the fused demand expression vector, including "card loss report expression vector", "accompanying medical treatment expression vector", and "power emergency repair expression vector"; E is the edge between services, used to represent service relationships, including "predecessor dependency", "similar services", and "parallel recommendation";

[0023] Calculate the service node similarity score between the semantic vector of the fused demand expression vector obtained in the multi-source vector fusion module and the service nodes V in the service knowledge graph G through the cosine similarity method; introduce the service node propagation influence factor, and calculate the preliminary node matching score based on the service node propagation influence factor and the service node similarity score;

[0024] Use the graph convolutional network GCN to learn the propagation relationship between the service nodes in the service knowledge graph G to obtain the attention weights between the service nodes in the service knowledge graph G , according to the attention weights between the service nodes And the preliminary node matching score are weighted and calculated to obtain the final node matching score;

[0025] Sort the service nodes in descending order according to the final node matching score, and generate a preliminary candidate service set based on the top K service nodes.

[0026] Preferably, in the candidate service optimization module, the three-layer natural language re-review model specifically includes:

[0027] A context intention reconstruction layer, which is used to process the preliminary candidate service set by using the multi-task BERT analysis method combined with embedded context prompt tags, output the semantic association score of the service nodes, re-rank the preliminary candidate service set according to the semantic association score, and output a first-layer optimized candidate service set;

[0028] A semantic redundancy disambiguation layer, which is used to perform semantic aggregation and exclusion on the first-layer optimized candidate service set, and output a second-layer optimized candidate service set;

[0029] A user emotion subjective preference adjustment layer, which is used to obtain historical satisfaction corpus, analyze the intention confidence according to the historical satisfaction corpus and the current resident portrait feature vector, and re-rank the second-layer optimized candidate service set according to the intention confidence, and output a third-layer optimized candidate service set.

[0030] Preferably, in the candidate service optimization module, the steps of sorting the preliminary candidate service set based on the three-layer natural language re-review model combined with the multi-objective sorting optimization method to generate an optimized candidate service set specifically include: obtaining a preferred recommendation score through multi-objective weighted calculation based on the first-layer optimized candidate service set, the second-layer optimized candidate service set, and the third-layer optimized candidate service set, and sorting the preliminary candidate service set according to the preferred recommendation score to generate an optimized candidate service set;

[0031] Among them, the formula used for calculating the preferred recommendation score is: ;

[0032] Among them, is the preferred recommendation score of the i-th service node ; is the priority score of the service node in the first-layer optimized candidate service set, which is determined according to the position of the service node in the first-layer optimized candidate service set; is the priority score of the service node in the second-layer optimized candidate service set, which is determined according to the position of the service node in the second-layer optimized candidate service set; is the priority score of the service node in the third-layer optimized candidate service set, which is determined according to the position of the service node in the third-layer optimized candidate service set; , and are target weighting coefficients, satisfying .

[0033] The present invention also provides an intelligent matching service method for community needs based on natural language processing, including:

[0034] Step S10: Obtain the original expression corpus of the current service window from the community server, perform OCR recognition and standardization processing on the original expression corpus to generate standard expression sentences; preset and train a natural language pre-training model, input the standard expression sentences into the natural language pre-training model, and output a text expression vector;

[0035] Step S20: Obtain the current resident portrait feature vector and the current service environment feature vector from the community server, and perform weighted fusion processing in combination with the text expression vector to obtain a fused demand expression vector;

[0036] Step S30: Preset a community service knowledge graph, perform semantic matching through the community service knowledge graph according to the fused demand expression vector, and generate a preliminary candidate service set;

[0037] Step S40: Set a three-layer natural language re-review model, and perform sorting processing on the preliminary candidate service set based on the three-layer natural language re-review model combined with the multi-objective sorting optimization method to generate an optimized candidate service set;

[0038] Step S50: Generate service recommendation cards according to the optimized candidate service set, and push the service recommendation cards to the resident side.

[0039] The present invention also provides a computer program product, including an intelligent matching service program for community needs based on natural language processing. When the intelligent matching service program for community needs based on natural language processing is executed by a processor, it implements the above-mentioned intelligent matching service method for community needs based on natural language processing.

[0040] The beneficial effects of the present invention are as follows: Compared with the existing community service recommendation methods that mostly rely on keyword or fixed template matching, especially under the conditions where the resident's expression is unclear or the context changes frequently, it is difficult to accurately identify service intentions and achieve intelligent recommendation. Since this application introduces a natural language pre-training model, constructs a community service knowledge graph and a multi-layer semantic re-review and sorting mechanism, it realizes semantic understanding of residents' service needs and personalized service push, thus avoiding problems such as large recommendation deviation, low service coverage, and poor response efficiency, and improving the intelligence and response accuracy of the community service system. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0042] Figure 1 It is a schematic diagram of the system of the first embodiment of an intelligent matching service system for community needs based on natural language processing according to the present invention.

[0043] Figure 2 It is a schematic diagram of the device of an intelligent matching service system for community needs based on natural language processing according to the present invention. Specific embodiments

[0044] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0045] Embodiment 1: As Figure 1 shown, it is a schematic flowchart of the first embodiment of the intelligent matching service system for community needs based on natural language processing according to the present invention, and the first embodiment of the intelligent matching service system for community needs based on natural language processing according to the present invention is proposed.

[0046] In the first embodiment, the intelligent matching service system for community needs based on natural language processing includes:

[0047] A text vector generation module, which is used to obtain the original expression corpus of the current service window from the community server, perform OCR recognition and standardization processing on the original expression corpus to generate standard expression sentences; preset and train a natural language pre-training model, input the standard expression sentences into the natural language pre-training model, and output text expression vectors;

[0048] It should be noted that in the text vector generation module, the original expression corpus specifically includes hotline voice transcription texts, community App messages, WeChat text and picture combination data, interview forms and visit records, mini-program search bar inputs, open-ended answers recognized by the AI outbound call system, handwritten requirement forms, medical records, and paper application forms. In the text vector generation module, the structure of the natural language pre-trained model specifically includes: an input encoding layer for encoding standard expression sentences into vector inputs that can be processed by the natural language pre-trained model; a multi-layer Transformer context modeling layer including a multi-head self-attention mechanism network, a feed-forward neural network, and a residual connection and layer normalization network; a fine-tuning mechanism processing layer for obtaining community-domain labeled corpus and supporting the use of the community-domain labeled corpus to perform lightweight fine-tuning on the semantic vectors output by the multi-layer Transformer context modeling layer; and an output layer for extracting the overall sentence semantic vector from the semantic vectors lightly fine-tuned and output by the fine-tuning mechanism processing layer to output the text expression vector.

[0049] It can be understood that the design of the text vector generation module breaks through the limitations of traditional keyword extraction or TF-IDF methods that can only process "canonical language" and "explicit keywords", and can automatically capture the implicit emotional expressions, role relationships, and service scenario contexts in sentences, and is applicable to resident service request texts with "incomplete expressions", "jumping statements", "vague words", and even "emotional colors".

[0050] It should be understood that to adapt to diverse expression scenarios in the community, the natural language pre-trained model enhances the model's semantic perception ability for local service expressions such as service terms through the domain corpus fine-tuning mechanism, and has stronger service-oriented semantic recognition accuracy and multi-round question-and-answer expression fusion ability than the general BERT. Compared with traditional static bag-of-words models or single-layer RNN models, the Transformer structure of this solution supports long-span dependency analysis.

[0051] For example, taking 1,000 pieces of original service expression data of residents collected by a social worker service platform in a certain city as an example, the traditional TF-IDF method, the LSTM + attention mechanism model, and the natural language pre-training model described in the present invention are respectively used to perform semantic matching tests on service items. The test finds that among the resident texts processed by the TF-IDF method, only about 524 texts are correctly matched in the Top-1 recommendation, and the accuracy rate is 52.4%; after using the deep model of LSTM + attention mechanism, the Top-1 accuracy rate is increased to 63.2%, but there are still certain recognition biases when facing colloquial and jumpy expressions. By using the natural language pre-training model in the present invention, through context modeling, domain fine-tuning, and multi-layer semantic integration mechanisms, the service matching ability can be significantly improved. Among the same 1,000 samples, 847 texts are accurately matched in the Top-1 recommendation, and the accuracy rate reaches 84.7%, reflecting good semantic generalization ability and actual implementation effect.

[0052] A multi-source vector fusion module is used to obtain the current resident portrait feature vector and the current service environment feature vector from the community service end, and perform weighted fusion processing in combination with the text expression vector to obtain a fusion requirement expression vector;

[0053] It should be noted that in the multi-source vector fusion module, the steps of obtaining the current resident portrait feature vector and the current service environment feature vector from the community service end, and performing weighted fusion processing in combination with the text expression vector to obtain a fusion requirement expression vector specifically include: obtaining the current resident portrait feature vector and the current service environment feature vector from the community service end; wherein, the current resident portrait feature vector is used to represent the current resident portrait; the current service environment feature vector is used to represent the environmental features of the current service, including time, channel, weather, and emergency status; first, perform preliminary weighted fusion on the text expression vector, the current resident portrait feature vector, and the current service environment feature vector through the three-vector weighted method to obtain a preliminary requirement expression vector; then use the deep neural network Transformer to further perform non-linear expression expansion processing on the preliminary requirement expression vector to obtain a fusion requirement expression vector.

[0054] It can be understood that in the present invention, the multi-source vector fusion module forms a more comprehensive requirement expression representation by weighted combination of the text expression vector, the resident portrait feature vector, and the service environment feature vector. Compared with the traditional method of only relying on text semantics for service matching, it more accurately reflects the service requirements of residents under specific population characteristics and environmental contexts. Especially when the resident's expression is unclear or ambiguous, the individual portrait information and environmental context can be used as important semantic complementation bases to accurately depict the true intention, effectively improving the accuracy and robustness of subsequent knowledge graph matching and re-review ranking.

[0055] It should be understood that the three-vector weighting only provides preliminary fusion expression ability. To further enhance the non-linear modeling effect of demand expression in the high-dimensional semantic space, the present invention introduces a deep Transformer structure on the basis of weighted fusion to expand the non-linear expression of the preliminary vectors. The multi-head attention mechanism in the Transformer structure can automatically capture the cross-dimensional dependence relationships in the fusion vectors, enabling the model to identify the true service intentions based on the combination of "resident category + environment type + language pattern". For example, for the same sentence "The card cannot be processed again", the system will tilt towards "assist in handling the loss report at the door" based on the context of "elderly living alone + voice channel + night expression", rather than recommending the default service item of "card loss report". This mechanism enhances the generalization ability and service matching sensitivity of the model in diverse input contexts.

[0056] For example, during the pilot deployment in a certain street, the system collected 500 resident service requests containing fuzzy expressions and compared the performance of three models in terms of service matching accuracy. The traditional BERT model (only based on text semantics) had a Top-1 recommendation accuracy of 68.2% for these requests; the accuracy of the model with the resident portrait information spliced as an additional label was improved to 75.4%; while after adopting the multi-source vector fusion module of "weighted fusion + Transformer non-linear expansion" described in the present invention, the Top-1 recommendation accuracy was improved to 88.6%. Among them, in the subsample of the "elderly group with weak expression ability", the accuracy improvement was the most significant, indicating that this solution has better service intention recognition ability for complex contexts and special populations.

[0057] A knowledge graph analysis module, configured to preset a community service knowledge graph, perform semantic matching through the community service knowledge graph according to the fused demand expression vector, and generate a preliminary candidate service set;

[0058] It should be noted that in the knowledge graph analysis module, the steps of presetting a community service knowledge graph and performing semantic matching through the community service knowledge graph according to the fused demand expression vector to generate a preliminary candidate service set specifically include: presetting a community service knowledge graph G, where G = (V, E); among them, V is a set of service nodes used to represent the fused demand expression vector, including "card loss reporting expression vector", "accompanying medical treatment expression vector", and "power emergency repair expression vector"; E is the edge between services used to represent service relationships, including "pre - dependency", "same - type service", and "parallel recommendation"; calculating the service node similarity score between the semantic vector of the fused demand expression vector obtained in the multi - source vector fusion module and the service nodes V in the service knowledge graph G through the cosine similarity method; introducing a service node propagation influence factor, and calculating a preliminary node matching score based on the service node propagation influence factor and the service node similarity score; using a graph convolutional network GCN to learn the propagation relationship between service nodes in the service knowledge graph G to obtain the attention weights between service nodes in the service knowledge graph G , according to the attention weights between service nodes and the preliminary node matching score, calculating the final node matching score through weighted calculation; sorting the service nodes in descending order according to the final node matching score, and generating a preliminary candidate service set based on the top K service nodes.

[0059] It can be understood that in the present invention, the knowledge graph analysis module realizes the accurate semantic matching between the fused demand expression vector and the service nodes by constructing a community service knowledge graph with structural relationship semantics. The service nodes in the knowledge graph not only carry semantic representation information but also express the logical dependencies, co - occurrence, and substitution relationships between services through edge structures, so that the structural context information can be effectively utilized during the matching process, avoiding the semantic island problem and improving the service recall ability under fuzzy expression conditions. By introducing the service node propagation influence and using the graph convolutional network to model the relationship weights between service nodes, the model not only focuses on the semantic distance of the current node itself but also can consider the surrounding semantically related services based on the graph structure expansion, realizing a comprehensive recommendation strategy of "semantic similarity + graph structure linkage".

[0060] It should be understood that most traditional service matching methods adopt static matching mechanisms based on keyword matching or sentence vector similarity, ignoring the upstream and downstream dependencies and structural coordination between services. Especially in community governance scenarios with fuzzy expressions or complex service structures, problems such as one - sided matching, misjudgment, or recommendation omission often occur. The present invention explicitly models the structural knowledge between service nodes by constructing a service knowledge graph, further introducing a graph propagation mechanism after the preliminary similarity calculation, and learning the structural influence between service nodes through a graph convolutional network to dynamically adjust the matching scores of nodes.

[0061] For example, in a simulation experiment, the system selected 300 service request texts expressed in natural language by residents. Approximately 28% of these expressions did not directly contain the names of standard service items. For example, "My mother is hospitalized and it is said that the card has not been paid and she cannot see a doctor". Under the traditional BERT semantic matching mechanism, the Top-1 service recommendation accuracy rate of such expressions was less than 55%. When the knowledge graph analysis module in the present invention was introduced, the system first screened out preliminary candidate nodes according to the semantic similarity between the fusion expression vector and the service node, and then propagated the importance through the graph convolutional network. Finally, the Top-1 recommended item was successfully matched. The overall results showed that in service expressions with structural dependencies, the Top-1 recommendation accuracy rate of the present invention increased by 29.3 percentage points, and the Top-3 recall rate reached more than 92%, significantly improving the coverage breadth and intention understanding depth of service recommendations.

[0062] A candidate service optimization module, configured to set a three-layer natural language re-review model, and perform sorting processing on the preliminary candidate service set based on the three-layer natural language re-review model in combination with a multi-objective sorting optimization method to generate an optimized candidate service set;

[0063] It should be noted that in the candidate service optimization module, the step of performing sorting processing on the preliminary candidate service set based on the three-layer natural language re-review model in combination with a multi-objective sorting optimization method to generate an optimized candidate service set specifically includes: obtaining a priority recommendation score through multi-objective weighted calculation based on the first-layer optimized candidate service set, the second-layer optimized candidate service set, and the third-layer optimized candidate service set, and performing sorting processing on the preliminary candidate service set according to the priority recommendation score to generate an optimized candidate service set; wherein, the formula used for calculating the priority recommendation score is: , where is the priority recommendation score of the i-th service node ; is the priority score of the service node in the first-layer optimized candidate service set, and is determined according to the position of the service node in the first-layer optimized candidate service set; is the priority score of the service node in the second-layer optimized candidate service set, and is determined according to the position of the service node in the second-layer optimized candidate service set; is the priority score of the service node in the third-layer optimized candidate service set, and is determined according to the position of the service node in the third-layer optimized candidate service set; , and are target weighting coefficients, satisfying .

[0064] It can be understood that in the present invention, the candidate service optimization module realizes a fusion evaluation mechanism for semantic matching, structural redundancy removal, and personalized preference by introducing three levels of optimization perspectives. The three-level optimization process focuses on the accuracy of expression semantics, the logical rationality between candidate services, and the subjective perception and historical behavior of residents, making the recommendation not only have language understanding ability but also reflect service adaptability and perceived satisfaction. Compared with the traditional single sorting method based only on semantic or rule scoring, this module has stronger intention review ability and recommendation sorting interpretability.

[0065] It should be understood that traditional recommendation sorting methods are often prone to distortion when dealing with scenarios of "incomplete expression" or "resident preference polarization". A single intention score may rank an option of "seemingly matching in expression but the service is not feasible" at the top, resulting in inaccurate recommendations. By introducing intention matching (the first layer), structural rationality screening (the second layer), and emotional preference adjustment (the third layer) into a unified sorting model and constructing a fusion sorting score function, the present invention effectively makes up for the problem that the scores of each layer are locally optimal but the overall recommendation is not good, and improves the global optimality and adaptability of service sorting.

[0066] For example, in a set of samples containing emotion-dominated expressions such as "The card is lost, the hospital refuses to see a doctor, and my mother is crying anxiously", a traditional model that only relies on intention matching is likely to take "Card loss reporting" as the top-1 recommended item, ignoring service items such as "Reimbursement assistance" or "Social worker accompanied medical treatment" that are more in line with the real perceived needs of the current user. After using the candidate service optimization module of the present invention: the first-layer score gives the highest score to "Card loss reporting"; in the second-layer score, it is downgraded due to an incomplete dependency chain; in the third-layer score, due to the strong emotion expressed by the user, the preference weights of service items such as "Accompanied medical treatment" and "Reimbursement assistance" are increased; in the final sorting result, "Card loss reporting" ranks second, and "Accompanied medical treatment" jumps to the first. The system recommendation is closer to the current perceived state of the user, and the accuracy of the top-1 recommendation has increased by more than 20%.

[0067] The service matching and pushing module is used to generate service recommendation cards according to the optimized candidate service set and push the service recommendation cards to the resident side.

[0068] It should be noted that by constructing a structured service recommendation card, the present invention realizes the conversion and expression of the internal intelligent sorting result of the system into a way that can be understood by users, improving the interpretability of the recommendation result and the user's trust. Combining the service matching reason and the handling guide, the recommendation card not only tells the user "what is recommended" but also explains "why it is recommended", effectively reducing the cognitive barrier of the elderly group to the system's automatic recommendation. At the same time, combined with the push scheduling mechanism, a closed-loop service experience of "delivering when the person arrives", "responding immediately to voice", and "handling immediately with the card" can be realized, improving the execution efficiency of the service recommendation system and the user reach ability.

[0069] It should be understood that the recommended information in traditional community service systems is often presented in the form of a list, lacking semantic interpretation and service guidance, which easily leads to users "not understanding, not daring to use, and not knowing what to do next", especially obvious in the elderly group or groups with weak expression ability. The present invention encapsulates the recommended service content through structured recommendation cards and adaptively matches them to different reach channels, taking into account the visibility, operability, and emergency recognition of recommendations, and overcomes the technical defects of the traditional recommendation system push method such as "single form, non-personalized push, and interactive break".

[0070] Embodiment 2: In addition, a method for intelligent matching service of community needs based on natural language processing provided by the present invention, using a system for intelligent matching service of community needs based on natural language processing in the above embodiment, can solve the technical problems of an intelligent matching service of community needs based on natural language processing. Compared with the prior art, the beneficial effects of a method for intelligent matching service of community needs based on natural language processing provided by the present invention are the same as those of a system for intelligent matching service of community needs based on natural language processing provided by the above embodiment, and other technical features in the method for intelligent matching service of community needs based on natural language processing are the same as the features disclosed in the method of the above embodiment, and will not be elaborated here.

[0071] Embodiment 3: The present invention provides a device for intelligent matching service of community needs based on natural language processing. Please refer to Figure 2, An intelligent matching service device for community needs based on natural language processing includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute a method for intelligent matching service of community needs in Embodiment 1 above. An intelligent matching service device for community needs in an embodiment of the present invention may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistant), PADs (Portable Application Description: tablet computers), PMPs (Portable Media Player: portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. An intelligent matching service device for community needs based on natural language processing is merely an example and should not impose any limitations on the functions and scope of use of the embodiments of the present invention. An intelligent matching service device for community needs based on natural language processing may include a processing device 1001 (such as a central processing unit, a graphics processing unit, etc.), which may perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM: Random Access Memory) 1004. In the RAM 1004, various programs and data required for the operation of an intelligent matching service device for community needs based on natural language processing are also stored. The processing device 1001, the ROM 1002, and the RAM 1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems may be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 may allow an intelligent matching service device for community needs based on natural language processing to communicate with other devices wirelessly or wiredly to exchange data. Although the figure shows an intelligent matching service device for community needs based on natural language processing having various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems may be alternatively implemented or had.

[0072] Embodiment 4: The present invention further provides a computer program product, including a computer program, and when the computer program is executed by a processor, it realizes the steps of a method for intelligent matching service of community requirements based on natural language processing as described above. The computer program product provided by the present invention can solve the technical problem of a method for intelligent matching service of community requirements based on natural language processing. Compared with the prior art, the beneficial effects of the computer program product provided by the present invention are the same as those of the method for intelligent matching service of community requirements based on natural language processing provided by the above embodiment, and will not be elaborated here.

[0073] Specifically, according to the embodiments disclosed by the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, an embodiment disclosed by the present invention includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program codes for executing the methods shown in the flowcharts. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by a processing device 1001, it executes the above functions defined in the methods of the embodiments disclosed by the present invention.

[0074] It should be understood that various parts disclosed by the present invention can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in a suitable manner in any one or more embodiments or examples.

[0075] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention also intends to include these changes and modifications.

Claims

1. An intelligent matching service system for community needs based on natural language processing, characterized in that, The system includes: A text vector generation module, which is used to obtain the original expression corpus of the current service window from the community server, perform OCR recognition and standardization processing on the original expression corpus to generate standard expression statements; preset and train a natural language pre-training model, input the standard expression statements into the natural language pre-training model, and output the text expression vectors; A multi-source vector fusion module, which is used to obtain the current resident portrait feature vector and the current service environment feature vector from the community server, and perform weighted fusion processing in combination with the text expression vector to obtain the fused demand expression vector; A knowledge graph analysis module, which is used to preset the community service knowledge graph, perform semantic matching through the community service knowledge graph according to the fused demand expression vector, and generate a preliminary candidate service set; A candidate service optimization module, which is used to set a three-layer natural language re-review model, and perform sorting processing on the preliminary candidate service set based on the three-layer natural language re-review model combined with the multi-objective sorting optimization method to generate an optimized candidate service set; A service matching and pushing module, which is used to generate service recommendation cards according to the optimized candidate service set and push the service recommendation cards to the resident side.

2. The intelligent matching service system for community needs based on natural language processing according to claim 1, characterized in that, In the text vector generation module, the original expression corpus specifically includes hotline voice transcription texts, community App messages, WeChat text and picture combination data, interview forms and visit records, mini-program search bar inputs, open-ended answers recognized by the AI outbound call system, handwritten demand forms, medical records, and paper application forms.

3. The intelligent matching service system for community needs based on natural language processing according to claim 1, wherein In the text vector generation module, the structure of the natural language pre-training model specifically includes: An input encoding layer, which is used to encode the standard expression statements into vector inputs that can be processed by the natural language pre-training model; A multi-layer Transformer context modeling layer, including a multi-head self-attention mechanism network, a feed-forward neural network, and a residual connection and layer normalization network; A fine-tuning mechanism processing layer, which is used to obtain community domain annotation corpora and support the use of community domain annotation corpora to perform lightweight fine-tuning on the semantic vectors output by the multi-layer Transformer context modeling layer; An output layer, which is used to extract the overall sentence semantic vector from the semantic vectors output by the lightweight fine-tuning of the fine-tuning mechanism processing layer and output the text expression vectors.

4. The intelligent matching service system for community needs based on natural language processing according to claim 1, characterized in that, In the multi-source vector fusion module, the steps of obtaining the current resident portrait feature vector and the current service environment feature vector from the community server and performing weighted fusion processing in combination with the text expression vector to obtain the fused demand expression vector specifically include: Obtain the current resident portrait feature vector and the current service environment feature vector from the community server; among them, the current resident portrait feature vector is used to represent the current resident portrait; the current service environment feature vector is used to represent the environmental features of the current service, including time, channel, weather, and emergency status; First, perform preliminary weighted fusion on the text expression vector, the current resident portrait feature vector, and the current service environment feature vector through the three-vector weighted method to obtain a preliminary demand expression vector; Then, use the deep neural network Transformer to further perform non-linear expression expansion processing on the preliminary demand expression vector to obtain the fused demand expression vector.

5. The intelligent matching service system for community needs based on natural language processing according to claim 1, characterized in that, In the knowledge graph analysis module, the steps of presetting a community service knowledge graph and performing semantic matching through the community service knowledge graph according to the fusion requirement expression vector to generate a preliminary candidate service set specifically include: Preset a community service knowledge graph G, where G = (V, E); among them, V is a set of service nodes for representing the fusion requirement expression vector, including "card loss reporting expression vector", "accompanying medical treatment expression vector", and "power emergency repair expression vector"; E is the edge between services for representing service relationships, including "pre - dependency", "same - type service", and "parallel recommendation". Calculate the service node similarity score between the semantic vector of the fusion requirement expression vector obtained in the multi - source vector fusion module and the service nodes V in the service knowledge graph G through the cosine similarity method; introduce a service node propagation influence factor, and calculate a preliminary node matching score based on the service node propagation influence factor and the service node similarity score. Use the graph convolutional network GCN to learn the propagation relationship between service nodes in the service knowledge graph G, and obtain the attention weights between service nodes in the service knowledge graph G , and calculate the final node matching score by weighted calculation according to the attention weights between service nodes and the preliminary node matching score; Sort the service nodes in descending order according to the final node matching score, and generate a preliminary candidate service set based on the top K service nodes.

6. The intelligent matching service system for community needs based on natural language processing according to claim 1, wherein In the candidate service optimization module, the three - layer natural language re - review model specifically includes: The context intention reconstruction layer is used to process the preliminary candidate service set using the multi - task BERT analysis method combined with embedded context prompt tags, output the semantic association score of the service nodes, re - sort the preliminary candidate service set according to the semantic association score, and output a first - layer optimized candidate service set. The semantic redundancy disambiguation layer is used to perform semantic aggregation and exclusion on the first - layer optimized candidate service set, and output a second - layer optimized candidate service set. The user emotion subjective preference adjustment layer is used to obtain historical satisfaction corpus, analyze the intention confidence according to the historical satisfaction corpus and the current resident portrait feature vector, and re - sort the second - layer optimized candidate service set according to the intention confidence, and output a third - layer optimized candidate service set.

7. The intelligent matching service system for community needs based on natural language processing according to claim 6, characterized in that, In the candidate service optimization module, the steps of sorting the preliminary candidate service set based on the three - layer natural language re - review model combined with the multi - objective sorting optimization method to generate an optimized candidate service set specifically include: obtaining a priority recommendation score through multi - objective weighted calculation based on the first - layer optimized candidate service set, the second - layer optimized candidate service set, and the third - layer optimized candidate service set, and sorting the preliminary candidate service set according to the priority recommendation score to generate an optimized candidate service set. Among them, the formula used for calculating the priority recommendation score is: ； Wherein, is the priority recommendation score of the i-th service node ; is the priority score of the service node in the first-layer optimized candidate service set, which is determined according to the position of the service node in the first-layer optimized candidate service set; is the priority score of the service node in the second-layer optimized candidate service set, which is determined according to the position of the service node in the second-layer optimized candidate service set; is the priority score of the service node in the third-layer optimized candidate service set, which is determined according to the position of the service node in the third-layer optimized candidate service set; , and are target weighting coefficients, satisfying .

8. A method for intelligent matching service of community requirements based on natural language processing, applied to a system for intelligent matching service of community requirements based on natural language processing according to any one of claims 1-7, characterized in that, The method includes: Step S10: Obtain the original expression corpus of the current service window from the community service end, perform OCR recognition and standardization processing on the original expression corpus to generate standard expression sentences; preset and train a natural language pre - training model, and input the standard expression sentences into the natural language pre - training model to output a text expression vector. Step S20: Obtain the current resident portrait feature vector and the current service environment feature vector from the community service end, and perform weighted fusion processing in combination with the text expression vector to obtain a fusion requirement expression vector. Step S30: Preset a community service knowledge graph, perform semantic matching through the community service knowledge graph based on the fusion requirement expression vector, and generate a preliminary candidate service set; Step S40: Set up a three-layer natural language re-review model, and perform sorting processing on the preliminary candidate service set based on the three-layer natural language re-review model combined with the multi-objective sorting optimization method to generate an optimized candidate service set; Step S50: Generate service recommendation cards according to the optimized candidate service set, and push the service recommendation cards to the resident side.

9. An intelligent matching service device for community needs based on natural language processing, characterized in that, The community demand intelligent matching service device based on natural language processing includes: a memory, a processor, and a community demand intelligent matching service program based on natural language processing stored on the memory and executable on the processor. When the community demand intelligent matching service program based on natural language processing is executed by the processor, it implements a community demand intelligent matching service system according to any one of claims 1 to 7.

10. A computer program product, characterized in that, The computer program product includes a community demand intelligent matching service program based on natural language processing. When the community demand intelligent matching service program based on natural language processing is executed by a processor, it implements a community demand intelligent matching service system according to any one of claims 1 to 7.

Images