Intelligent thermal work order system based on AI algorithm data processing

Abstract

The present invention discloses an intelligent thermal work order system based on AI algorithm data processing, which specifically relates to the field of place name mapping and conflict handling, and is used to solve the spatial conflict and accuracy problems in matching dialect place names with standard place names. It collects and analyzes the implicit association between place name descriptions in historical work orders and actual work order coordinates, and combines public GIS data, community text information and local databases to automatically process and update place name mapping. When faced with place names with dialect characteristics or outdated place names, not only can effective matching be performed, but also the place name mapping dictionary can be continuously updated and optimized online through machine learning and natural language processing technology; in this way, the mapping mechanism can adapt to urban development, language changes and changes in user habits in real time, ensuring that the work order system can accurately and quickly identify and locate place names, thereby effectively improving scheduling efficiency and the accuracy of resource allocation.

Description

Technical Field

[0001] The present invention relates to the field of place name mapping and conflict resolution, and more specifically, to an intelligent thermal work order system based on AI algorithm data processing. Background Art

[0002] In the thermal work order processing process, the work order address information is the key basis for dispatching personnel and resource allocation. However, customers or housekeepers often use non-standardized geographical references during calls, such as old place names from many years ago, habitual names of local residents, and abandoned or merged administrative division names. Such place names cannot be directly matched to standard coordinates or administrative divisions in traditional address resolution libraries, resulting in the system being unable to effectively locate addresses, assign tasks, or conduct subsequent analysis and decision-making. In particular, with the evolution of urban planning and changes in residents' language, non-standardized dialect place names continue to emerge. Solving this problem requires not only overcoming the limitations of static rules, but also dynamic and adaptive update capabilities.

[0003] The core problem facing the current thermal work order system is the lack of a flexible and intelligent place name mapping mechanism. In the face of non-standard situations such as dialect place names, accent variations, and historical place names, traditional address parsing libraries can only determine the location through strict matching. When a match fails, it usually remains in a failed state, resulting in subsequent scheduling and analysis not being able to proceed smoothly. The root cause of this problem is that the existing system relies on static rules and dictionaries, and lacks a mechanism for dynamic updates and real-time optimization. Summary of the invention

[0004] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides an intelligent thermal work order system based on AI algorithm data processing, which automatically processes and updates the place name mapping by collecting and analyzing the implicit association between the place name description in the historical work order and the actual work order coordinates, combined with public GIS data, community text information and local databases. When faced with place names with dialect characteristics or outdated place names, not only can effective matching be performed, but also the place name mapping dictionary can be continuously updated and optimized online through machine learning and natural language processing technology; in this way, the mapping mechanism can adapt to urban development, language changes and changes in user habits in real time, ensuring that the work order system can accurately and quickly identify and locate place names, thereby effectively improving scheduling efficiency and the accuracy of resource allocation, so as to solve the problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] Intelligent thermal work order system based on AI algorithm data processing, including: voice transcription module, place name recognition module, dynamic mapping module and geographic verification module;

[0007] Speech transcription module: performs speech-to-text processing on call recordings, marks potential key indicators, and outputs the transcribed text data to the place name recognition module;

[0008] Place name recognition module: Based on the transcribed text data, it identifies dialect place names that cannot be matched in the standard address library, generates a dialect place name candidate list, and outputs the dialect place name candidate list to the dynamic mapping module;

[0009] Dynamic mapping module: Use historical work order data, local community text data and public GIS information to perform semantic clustering and similarity calculation on the candidate list of dialect place names, determine the mapping relationship between dialect place names and standard place names, and add the mapping relationship to the place name mapping dictionary; output the place name mapping results to the geographic verification module;

[0010] Geographic verification module: perform spatial verification on the place name mapping results, verify the road structure, regional boundaries and building distribution through public GIS information to ensure that the mapping relationship is reasonable; if there is a conflict, select the next optimal mapping.

[0011] In a preferred embodiment, the processing process of the speech transcription module includes the following:

[0012] First, pre-process the recording of the call between the butler and the customer to ensure the clarity of the voice signal;

[0013] Based on the clear speech signal, deep learning algorithms are used to transcribe speech to text. By comparing the speech recognition model trained on a large-scale annotated corpus, each audio segment is transcribed and preliminary text data is output.

[0014] Using a deep learning-based text analysis model, we perform fine-grained word segmentation on the transcribed text to separate key vocabulary units. Then, we combine the domain knowledge base to identify and mark the key information in the text.

[0015] For each key information extracted from the text, the context analysis module is used to determine its relevance to the entire call content;

[0016] Finally, each key information in the transcribed text is clearly marked and associated with its context.

[0017] In a preferred embodiment, the processing of the place name recognition module includes the following:

[0018] Use Chinese word segmentation technology to split the transcribed text into sentences and words; annotate place name words through training data to identify potential boundaries of place names;

[0019] By combining rule-based place name recognition with machine learning, we extract dialect place names from transcribed texts. For place names that cannot be directly matched in the standard address database, we use the context of the words in the text to make inferences. We use language models to perform contextual reasoning to capture the semantic connection between place names and other words.

[0020] Based on the candidate place names extracted from the transcribed text, historical work order data is called for semantic understanding and the similarity between the candidate place names and the standard place names is calculated;

[0021] By calling the standard address library, candidate place names are matched and screened;

[0022] For dialect place names that cannot be directly matched, a fuzzy matching algorithm is used to combine community text data and geographic information for secondary optimization. At this time, candidate place names will be calculated based on geographic location and historical records, and the place names that best match the context will be selected through a similarity algorithm.

[0023] Finally, an optimized candidate list of dialect place names is generated.

[0024] In a preferred embodiment, the processing of the dynamic mapping module includes the following:

[0025] Extract historical work order data from the internal database; Analyze the correspondence between place name descriptions in historical work order data and actual dispatch locations, and build a preliminary mapping reference between dialect place names and standard place names;

[0026] Use crawler technology to collect text data containing place name descriptions from local community forums, social media platforms, and local news websites; use natural language processing technology to clean and filter out potential dialect place names and their contextual information to form a place name corpus;

[0027] Access the public GIS database to obtain standard place names, geographic coordinates, administrative division information and geographic feature data;

[0028] Using the pre-trained BERT model, each place name and its context sentence in the dialect place name candidate list is converted into a high-dimensional semantic embedding vector;

[0029] A density-based clustering algorithm is used to perform cluster analysis on semantic embedding vectors to identify place name groups with similar semantic features.

[0030] In a preferred embodiment, the processing of the dynamic mapping module further includes the following:

[0031] A comprehensive similarity calculation model is constructed, combining multiple similarity indicators to evaluate the similarity between dialect place names and standard place names in multiple dimensions; the specific calculation formula is as follows: ;in: Indicates Dialect place names and The comprehensive similarity score between the standard place names; and Semantic embedding vectors representing dialect place names and standard place names respectively; and The phonetic or lexical collections representing dialect place names and standard place names respectively; , , is the weight coefficient;

[0032] Using genetic algorithm to optimize weight coefficients , , , in order to maximize the proportion of correct matches in historical work order data; the specific optimization objective function is as follows: ;in: Indicates the total number of historical work order data; For the A binary variable indicating whether a work order is matched successfully or not; Indicates The similarity score between the dialect place name and the standard place name in each work order;

[0033] Based on the comprehensive similarity score, the standard place name with the highest score is selected as the mapping relationship for each dialect place name; if the highest score is lower than the preset threshold, it is marked as awaiting manual review; for the place name relationship that is automatically mapped successfully, it is added to the place name mapping dictionary; for the place names awaiting manual review, they are submitted to the manual review module, the final mapping relationship is determined through manual intervention, and the dictionary is subsequently updated.

[0034] In a preferred embodiment, the processing of the geographic verification module includes the following:

[0035] Each mapping result needs to be compared with the geographic coordinates of the standard place name during the preliminary verification; by comparing the preliminary mapping coordinates of the dialect place name with the known spatial coordinates of the standard place name, the following spatial comparison formula is executed: ;in: Indicates Dialect place names and The spatial distance between standard place names; , Indicates dialect place names The mapping coordinates of , Indicates a standard place name The actual coordinates of Represents the normalized radius of geographic coordinates;

[0036] If the spatial distance is less than the preset threshold, the corresponding place name mapping is considered reasonable and the next step of verification is entered.

[0037] In a preferred embodiment, the processing of the geographic verification module also includes the following:

[0038] By comparing the regional boundary data and building distribution, boundary overlap calculation is performed to ensure that the place name mapping conforms to the actual regional division: ;in:

[0039] Indicates Dialect place names and The degree of boundary overlap of standard place names on regional boundaries; and They respectively represent the geographical boundaries of the areas corresponding to dialect place names and standard place names; if the boundary overlap is greater than the set tolerance, it is considered a conflict and enters the conflict resolution stage.

[0040] In a preferred embodiment, the processing of the geographic verification module also includes the following:

[0041] When there are spatial conflicts among multiple place name mappings, the optimal mapping needs to be selected. The following comprehensive evaluation function is used for optimal selection: ;in: Indicates Dialect place names and The comprehensive matching score of the standard place names; is a function of various influencing factors; is the coefficient of each factor; , They are spatial distance and boundary overlap respectively; by adjusting the coefficients of each factor, the most appropriate mapping relationship can be automatically selected and incorrect mapping due to conflicts can be avoided;

[0042] After the optimal mapping selection is completed, the conflicting items are cleared, the place names that meet the conflict criteria are re-marked for manual review, and new mapping candidate relationships are generated to enter the next round of spatial verification and conflict resolution.

[0043] The technical effects and advantages of the intelligent thermal work order system based on AI algorithm data processing of the present invention are as follows:

[0044] By collecting and analyzing the implicit association between place name descriptions in historical work orders and actual work order coordinates, combined with public GIS data, community text information and local databases, place name mapping is automatically processed and updated. When faced with place names with dialect characteristics or outdated place names, not only can effective matching be performed, but the place name mapping dictionary can also be continuously updated and optimized online through machine learning and natural language processing technology; in this way, the mapping mechanism can adapt to urban development, language changes and changes in user habits in real time, ensuring that the work order system can accurately and quickly identify and locate place names, thereby effectively improving scheduling efficiency and the accuracy of resource allocation. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a schematic diagram of the structure of the intelligent thermal work order system based on AI algorithm data processing of the present invention. DETAILED DESCRIPTION

[0046] 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 only 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.

[0047] Embodiment 1: Figure 1 The present invention provides an intelligent thermal work order system based on AI algorithm data processing, including: a speech transcription module, a place name recognition module, a dynamic mapping module and a geographic verification module.

[0048] Speech transcription module: performs speech-to-text processing on call recordings, marking potential key indicators; outputs the transcribed text data to the place name recognition module.

[0049] Place name recognition module: Based on the transcribed text data, it identifies dialect place names that cannot be matched in the standard address library, generates a candidate list of dialect place names, and outputs the list to the dynamic mapping module.

[0050] Dynamic mapping module: Utilizes historical work order data, local community text data and public GIS information to perform semantic clustering and similarity calculation on the candidate list of dialect place names, determine the mapping relationship between dialect place names and standard place names, and add the mapping relationship to the place name mapping dictionary; outputs the place name mapping results to the geographic verification module.

[0051] Geographic verification module: perform spatial verification on the place name mapping results, verify the road structure, regional boundaries and building distribution through public GIS information to ensure that the mapping relationship is reasonable; if there is a conflict, select the next optimal mapping.

[0052] In the intelligent thermal work order system, the core task of the voice transcription module is to accurately convert the conversation recordings between the housekeeper and the customer into text data, and mark key indicator words during the transcription process to provide accurate information support for subsequent place name recognition and work order processing. Voice transcription must not only take into account interference factors such as dialects, accents, and noise, but also use intelligent algorithms to accurately capture and mark information that is critical to work order processing, such as fault type, location description, customer emotions, etc. This step is crucial for subsequent place name recognition, fault assessment, and dispatch scheduling. Accurate voice transcription results directly affect the efficiency and accuracy of the entire work order processing.

[0053] The processing of the speech transcription module includes the following:

[0054] 1.1. First, pre-process the call recordings between the butler and the customer, including noise removal, speech speed balance and voice signal enhancement processing to ensure the clarity of the voice signal. The recordings are optimized by using signal processing algorithms (such as short-time Fourier transform and linear predictive coding) to filter out background noise, echo and other irrelevant sounds.

[0055] 1.2. Based on the clear voice signal, deep learning algorithms (such as acoustic models based on recurrent neural networks (RNNs) and Transformer models) are used to transcribe speech to text. By comparing the speech recognition model trained on a large-scale annotated corpus, each audio segment is transcribed and preliminary text data is output. At this time, the speech recognition model needs to have high-precision dialect recognition capabilities and professional vocabulary libraries in specific fields (such as thermal systems, home appliance repairs, etc.) to ensure that the converted text is as close as possible to the customer's intention.

[0056] 1.3. After generating the preliminary transcription text, natural language processing (NLP) technology is used to perform text segmentation and syntactic analysis. Using a deep learning-based text analysis model, the transcription text is segmented at a fine-grained level to separate key vocabulary units. Then, combined with the domain knowledge base, key information in the text is identified and marked, such as fault type (such as insufficient heating, pipeline leakage), geographic location (such as place name, direction), time, and emotional attitude (such as customer anxiety, urgency). This key information will serve as the basis for subsequent work order processing and dispatching decisions.

[0057] 1.4. For each key information extracted from the text, the context analysis module is used to determine its relevance and importance to the entire call content. This process uses semantic analysis algorithms, such as the BERT-based context understanding model, to ensure that accurate intent is captured in more complex conversations. By analyzing the semantic dependency of the previous and next sentences, misunderstandings of ambiguous words or vague descriptions are avoided, ensuring that each key information point can be effectively matched with subsequent work order processing.

[0058] 1.5. Finally, after a series of processing, each key information in the transcribed text (such as place names, fault descriptions, customer emotions, etc.) will be clearly labeled and associated with its context. All transcribed data will form a structured output and be passed to the place name recognition module and the fault assessment module for subsequent processing. The output data format is unified into text with key information tags to ensure that subsequent modules can seamlessly connect and correctly extract the required data.

[0059] The speech transcription module converts unstructured voice information into structured text data through efficient processing of call recordings, and performs semantic annotation to provide accurate data support for subsequent place name recognition, fault analysis and work order distribution. This process involves multiple processing steps, from signal optimization, speech recognition to text analysis and indicator word extraction, and each step relies on the close cooperation of deep learning algorithms and NLP technology. Through this modular processing method, it can ensure that even in voice data with complex situations such as dialects, terminology, and noise, customer needs and fault characteristics can be accurately captured, thereby providing reliable data support for the intelligent thermal work order system.

[0060] In the intelligent thermal work order system, the place name recognition module plays a vital role following the text data output of the speech transcription module. Through this module, the system can automatically identify dialectal place names and common landmark information that is difficult to match the standard address library from the transcribed text. The accuracy of place name recognition is crucial to the efficient operation of the entire work order processing process, especially in a multi-dialect and multi-region environment. In order to ensure that under complex and diverse geographical expressions, the place name recognition module can accurately extract potential place name information and generate a candidate list for further processing by subsequent modules.

[0061] The processing of the place name recognition module includes the following:

[0062] 2.1-1, the input data of the place name recognition module is the transcribed text data output by the speech transcription module. This text may contain dialectal place names, common names, pinyin misrecognition, grammatical ambiguity and other problems. The module first cleans the data to remove non-place name noise information (such as irrelevant words, stop words, etc.) to ensure the cleanliness and effectiveness of the processed data.

[0063] 2.1-2, use Chinese word segmentation technology (such as CRF, BERT embedding word segmentation model) to split the transcribed text into sentences and words. In the word segmentation process, the system pays special attention to place-related nouns, phrases, and possible compound place names (such as "Nancheng East Road" or "Laodongmen"). By annotating place name words with training data, the system can identify the potential boundaries of place names.

[0064] 2.2, through the combination of rule-based place name recognition and machine learning, the system can extract dialect place names from transcribed texts. For place names that cannot be directly matched in the standard address library (such as local dialects and common names), the system infers through the context of the words in the text. Language models (such as GPT and BERT) are used for contextual reasoning to capture the semantic connection between place names and other words.

[0065] 2.3, based on the candidate place names extracted from the transcribed text, the system calls the historical work order data for semantic understanding and calculates the similarity between the candidate place names and the standard place names. Here, a semantic similarity calculation model based on BERT is used, which can understand the distance between the semantics of dialect place names and standard place names.

[0066] 2.4-1, the system accurately matches and screens candidate place names by calling standard address libraries (such as public GIS information and industry place name databases). At this time, the match does not only rely on word similarity, but also combines the historical frequency of place names and the relevance of geographical locations. For those candidates with low frequency of place names or large geographical spans, the system will give a lower matching weight to exclude possible wrong matches.

[0067] 2.4-2, for dialect place names that cannot be directly matched, the system uses a fuzzy matching algorithm, combined with community text data and geographic information, for secondary optimization. At this time, the candidate place names will be further calculated based on dimensions such as geographic location and historical records, and the place names that best match the context will be selected through a similarity algorithm.

[0068] 2.5-1, Finally, the place name recognition module will generate an optimized dialect place name candidate list based on all the above steps. The list contains multiple standard place name candidates, each with its similarity, historical matching records and possible geographical location information. This list will be transmitted as input data to the dynamic mapping module.

[0069] 2.5-2, the candidate place name list is output in the form of structured data (such as JSON) and transmitted to the downstream module through the API interface. The data contains relevant information of the candidate place names, such as place name text, similarity value, corresponding historical records, etc., so that the dynamic mapping module can perform subsequent place name standardization and spatial verification.

[0070] The place name recognition module in the intelligent thermal work order system is responsible for extracting and optimizing the candidate list of dialect place names from the transcribed text. Through a variety of technical means, including text segmentation, dialect place name recognition, semantic understanding and similarity calculation, the system can accurately identify dialect place names and their potential standardized names. By combining with historical work order data, public GIS information and contextual context, the module can effectively optimize dialect place names to ensure the accuracy and rationality of place names. This module provides strong support for the subsequent dynamic mapping and geographic verification modules, ensuring the efficient operation of the intelligent thermal work order system.

[0071] The dynamic mapping module undertakes the core task of mapping the dialect place name candidate list generated by the place name recognition module to standard place names. This module uses advanced semantic clustering and similarity calculation technology to accurately determine the correspondence between dialect place names and standard place names by comprehensively utilizing historical work order data, local community text data, and public geographic information system (GIS) data. The efficient operation of the dynamic mapping module not only improves the accuracy of place name resolution, but also enhances the system's adaptability to emerging dialect place names and geographical changes, thereby ensuring that subsequent geographic verification and fault assessment modules can be efficiently processed based on accurate geographic information.

[0072] The processing of the dynamic mapping module includes the following:

[0073] 3.1-1, first extract historical work order data from the internal database, especially those records that contain dialect place name descriptions and have been successfully dispatched. By parsing the correspondence between the place name descriptions of historical work order data and the actual dispatch locations, a preliminary mapping reference between dialect place names and standard place names is constructed.

[0074] 3.1-2, using crawler technology to collect text data containing place name descriptions from local community forums, social media platforms and local news websites. Using natural language processing technology to clean and filter out potential dialect place names and their contextual information, a rich place name corpus is formed.

[0075] 3.1-3, access the public GIS database to obtain standard place names, geographic coordinates, administrative division information and geographic feature data (such as road network, building distribution, etc.). Regularly update GIS data through the API interface to ensure the timeliness and accuracy of place name mapping.

[0076] 3.2-1, using the pre-trained BERT model, each place name and its context sentence in the dialect place name candidate list is converted into a high-dimensional semantic embedding vector. The specific formula is as follows: ;

[0077] in:

[0078] Indicates Semantic embedding vectors of dialect place names.

[0079] A sentence indicating the context of a place name in a text.

[0080] Indicates A dialect place name.

[0081] 3.2-2, use density-based clustering algorithms (such as DBSCAN) to cluster semantic embedding vectors and identify place name groups with similar semantic features. The clustering process adjusts the density threshold and minimum sample number parameters to adapt to the distribution characteristics of place names in different dialects.

[0082] ;

[0083] in:

[0084] Indicates Clustering results.

[0085] It is the neighborhood radius parameter, which determines the density requirement of clustering.

[0086] is the minimum sample number parameter, which determines the minimum number of points required to form a cluster.

[0087] 3.3-1, build a comprehensive similarity calculation model, combining multiple similarity indicators (such as cosine similarity , Euclidean distance Similarity to Jaccard ), which evaluates the similarity between dialect place names and standard place names in multiple dimensions. The specific calculation formula is as follows: ;

[0088] in:

[0089] Indicates Dialect place names and The comprehensive similarity score between the standard place names.

[0090] and Represent the semantic embedding vectors of dialect place names and standard place names respectively.

[0091] and A collection of pinyin or words that represent dialect place names and standard place names respectively.

[0092] , , is the weight coefficient, which reflects the importance of each similarity index in the comprehensive score.

[0093] 3.3-2, using genetic algorithm (GA) to optimize weight coefficients , , , in order to maximize the proportion of correct matches in historical work order data. The specific optimization objective function is as follows: ;

[0094] in:

[0095] Indicates the total number of historical work order data.

[0096] For the A binary variable indicating whether a work order is matched successfully or not (1 for success, 0 for failure).

[0097] Indicates The similarity score between the dialect place name and the standard place name in each work order.

[0098] 3.4-1, based on the comprehensive similarity score, select the standard place name with the highest score for each dialect place name as its mapping relationship. If the highest score is lower than the preset threshold, it will be marked for manual review.

[0099] ;

[0100] in:

[0101] Indicates The standard place name mapping results of dialect place names.

[0102] It is the similarity threshold, which determines whether automatic mapping or manual review is required.

[0103] 3.4-2, for the place name relationships that are successfully mapped automatically, add them to the place name mapping dictionary to ensure that the same or similar place names can be quickly identified and matched in the future. For the place names to be manually reviewed, submit them to the manual review module, determine the final mapping relationship through manual intervention, and then update the dictionary.

[0104] 3.4-3, through the feedback mechanism, the accuracy of the mapping results will be compared with the verification results of the subsequent geographic verification module. For incorrect or unreasonable mapping results, the system will adjust the similarity calculation model or retrain the clustering algorithm to improve the accuracy of the mapping.

[0105] After completing the mapping relationship determination and verification, the system will output the standardized place name mapping results to the geographic verification module in structured data (such as JSON format).

[0106] The dynamic mapping module integrates historical work order data, local community text data and public GIS information, and uses advanced semantic clustering and multi-dimensional similarity calculation technology to accurately map dialect place names to standard place names. The semantic embedding, clustering analysis, genetic algorithm optimization and other technical means within the module ensure the high accuracy and adaptability of place name mapping. At the same time, the dynamically updated place name mapping dictionary and feedback mechanism enable the system to continuously learn and optimize to adapt to the ever-changing place name expressions and geographical environment. The efficient operation of the dynamic mapping module provides reliable geographic information support for subsequent geographic verification and fault assessment modules, greatly improving the overall intelligence level and service quality of the intelligent thermal work order system.

[0107] In the intelligent thermal work order system, the geographic verification module is a key component to ensure that the place name mapping results conform to the actual geographic spatial relationship. By performing spatial verification on the place name mapping results, combined with the support of public GIS data and geographic information systems, the geographic verification module ensures the accuracy of each place name mapping and the actual geographic location. This process not only verifies the rationality of the place name and spatial coordinates, but also identifies possible geographic conflicts, provides the optimal geographic coordinates and standard administrative division information, and thus provides accurate input data for the fault assessment module.

[0108] The geographic verification module ensures the accuracy and rationality of the place name mapping results output by the dynamic mapping module through precise spatial comparison, boundary overlap analysis and conflict handling. Its technologies include spatial distance calculation, regional conflict detection and comprehensive evaluation model, combined with historical work order data and GIS information to ensure that each place name mapping is consistent with the actual geographical location. The module is continuously optimized through the feedback mechanism to adapt to new geographical features and dialect expressions, and ultimately provides accurate geographical coordinates and administrative division information for the fault assessment module, thereby improving the intelligence and response efficiency of the entire system.

[0109] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0110] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

[0111] It should be noted that, in this article, if there are relational terms such as first and second, etc., they are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "including a..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0112] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. Intelligent thermal work order system based on AI algorithm data processing, characterized by: include: Voice transcription module, place name recognition module, dynamic mapping module and geographic verification module; Speech transcription module: performs speech-to-text processing on call recordings, marks potential key indicators, and outputs the transcribed text data to the place name recognition module; Place name recognition module: Based on the transcribed text data, it identifies dialect place names that cannot be matched in the standard address library, generates a dialect place name candidate list, and outputs the dialect place name candidate list to the dynamic mapping module; Dynamic mapping module: using historical work order data, local community text data and public GIS information, to perform semantic clustering and similarity calculation on the candidate list of dialect place names, determine the mapping relationship between dialect place names and standard place names, and add the mapping relationship to the place name mapping dictionary; output the place name mapping results to the geographic verification module; the processing process of the dynamic mapping module includes the following: Extract historical work order data from the internal database; Analyze the correspondence between place name descriptions in historical work order data and actual dispatch locations, and build a preliminary mapping reference between dialect place names and standard place names; Use crawler technology to collect text data containing place name descriptions from local community forums, social media platforms, and local news websites; use natural language processing technology to clean and filter out potential dialect place names and their contextual information to form a place name corpus; Access the public GIS database to obtain standard place names, geographic coordinates, administrative division information and geographic feature data; Using the pre-trained BERT model, each place name and its context sentence in the dialect place name candidate list is converted into a high-dimensional semantic embedding vector; A density-based clustering algorithm is used to cluster the semantic embedding vectors to identify place name groups with similar semantic features. Construct a comprehensive similarity calculation model that combines multiple similarity metrics, including cosine similarity , Euclidean distance Similarity to Jaccard , evaluate the similarity between dialect place names and standard place names in multiple dimensions; the specific calculation formula is as follows: ;in: Indicates Dialect place names and The comprehensive similarity score between the standard place names; and Semantic embedding vectors representing dialect place names and standard place names respectively; and The phonetic or lexical collections representing dialect place names and standard place names respectively; , , They are , , The weight coefficient of Using genetic algorithm to optimize weight coefficients , , , in order to maximize the proportion of correct matches in historical work order data; the specific optimization objective function is as follows: ;in: Indicates the total number of historical work order data; For the A binary variable indicating whether a work order is matched successfully or not; Indicates The similarity score between the dialect place name and the standard place name in each work order; According to the comprehensive similarity score, the standard place name with the highest score is selected as the mapping relationship for each dialect place name; if the highest score is lower than the preset threshold, it is marked as waiting for manual review; for the place name relationship that is automatically mapped successfully, it is added to the place name mapping dictionary; for the place names waiting for manual review, it is submitted to the manual review module, and the final mapping relationship is determined through manual intervention, and then the dictionary is updated; Geographic verification module: perform spatial verification on the place name mapping results, verify the road structure, regional boundaries and building distribution through public GIS information to ensure that the mapping relationship is reasonable; if there is a conflict, select the next optimal mapping.

2. The intelligent thermal work order system based on AI algorithm data processing according to claim 1 is characterized in that: The processing of the speech transcription module includes the following: First, pre-process the recording of the call between the butler and the customer to ensure the clarity of the voice signal; Based on the clear voice signal, deep learning algorithm is used to transcribe speech to text; By comparing the speech recognition model trained on a large-scale annotated corpus, each audio segment is transcribed and preliminary text data is output; Using a deep learning-based text analysis model, we perform fine-grained word segmentation on the transcribed text to separate key vocabulary units. Then, we combine the domain knowledge base to identify and mark the key information in the text. For each key information extracted from the text, the context analysis module is used to determine its relevance to the entire call content; Finally, each key information in the transcribed text is clearly marked and associated with its context.

3. The intelligent thermal work order system based on AI algorithm data processing according to claim 2 is characterized in that: The processing of the place name recognition module includes the following: Use Chinese word segmentation technology to split the transcribed text into sentences and words; use training data to annotate place name words and identify potential boundaries of place names; By combining rule-based place name recognition with machine learning, we extract dialect place names from transcribed texts. For place names that cannot be directly matched in the standard address database, we use the context of the words in the text to make inferences. We use language models to perform contextual reasoning to capture the semantic connection between place names and other words. Based on the candidate place names extracted from the transcribed text, historical work order data is called for semantic understanding and the similarity between the candidate place names and the standard place names is calculated; By calling the standard address library, candidate place names are matched and screened; For dialect place names that cannot be directly matched, a fuzzy matching algorithm is used to combine community text data and geographic information for secondary optimization. At this time, candidate place names will be calculated based on geographic location and historical records, and the place names that best match the context will be selected through a similarity algorithm. Finally, an optimized candidate list of dialect place names is generated.

4. The intelligent thermal work order system based on AI algorithm data processing according to claim 1 is characterized in that: The processing of the geographic verification module includes the following: Each mapping result needs to be compared with the geographic coordinates of the standard place name during the preliminary verification; by comparing the preliminary mapping coordinates of the dialect place name with the known spatial coordinates of the standard place name, the following spatial comparison formula is executed: ;in: Indicates Dialect place names and The spatial distance between standard place names; , Indicates dialect place names The mapping coordinates of , Indicates a standard place name The actual coordinates of Represents the normalized radius of geographic coordinates; If the spatial distance is less than the preset threshold, the corresponding place name mapping is considered reasonable and the next step of verification is entered.

5. The intelligent thermal work order system based on AI algorithm data processing according to claim 4 is characterized in that: The processing of the geographic verification module also includes the following: By comparing the regional boundary data and building distribution, boundary overlap calculation is performed to ensure that the place name mapping conforms to the actual regional division: ;in: Indicates Dialect place names and The degree of boundary overlap of standard place names on regional boundaries; and They respectively represent the geographical boundaries of the areas corresponding to dialect place names and standard place names; if the boundary overlap is greater than the set tolerance, it is considered a conflict and enters the conflict resolution stage.

6. The intelligent thermal work order system based on AI algorithm data processing according to claim 5 is characterized in that: The processing of the geographic verification module also includes the following: When there are spatial conflicts among multiple place name mappings, the optimal mapping needs to be selected. The following comprehensive evaluation function is used for optimal selection: ;in: Indicates Dialect place names and The comprehensive matching score of the standard place names; is a function of various influencing factors; is the coefficient of each factor; , They are spatial distance and boundary overlap respectively; by adjusting the coefficients of each factor, the most appropriate mapping relationship can be automatically selected and incorrect mapping due to conflicts can be avoided; After the optimal mapping selection is completed, the conflicting items are cleared, the place names that meet the conflict criteria are re-marked for manual review, and new mapping candidate relationships are generated to enter the next round of spatial verification and conflict resolution.

Images