Intelligent urban service guide method and system

Abstract

The application discloses a kind of urban service intelligent guide methods and systems, it is related to data analysis technical field, the method includes: the service demand description information of user input is preprocessed;Based on the service demand description information, demand semantic diagram is constructed to utilize city service knowledge graph to carry out intention recognition;Based on the service demand description information, city service knowledge graph is used to determine disambiguation mapping table to carry out professional word matching;Based on target intention information and target professional word, determine pre-incident link to carry out the logical reasoning of the type of urban service affair, obtain affair dependency relationship graph;Based on affair dependency relationship graph, utilize city digital twin service space to carry out the guide path analysis of urban service to execute the urban service guide of user.The application improves the intelligent degree of service guide, so that user obtains more satisfactory use experience.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, and in particular to a smart guidance method and system for urban services. Background Technology

[0002] With the increasing number of urban public services, citizens often face the practical dilemma of not knowing where to go or what to do first when handling various urban services. Therefore, intelligent guidance technology is of great significance for improving the efficiency of government services and user experience. However, existing urban service guidance methods still fall short of actual needs. On the one hand, the service requests input by users are mostly colloquial and fragmented natural language. Existing methods mostly only use shallow intent classification, which is difficult to accurately identify the complete intent from a deeper semantic level, leading to misjudgment of intent and a lack of accurate semantic anchors for subsequent guidance. On the other hand, real-world urban service processing generally involves strict pre- and post-processing dependencies. For example, real estate registration often requires completing pre-processing such as paying taxes and obtaining a family housing certificate. Existing methods generally lack the ability to deeply reason about the logic of the process and do not integrate business dependencies with real-world service space information. Users still need to sort out the process chain themselves, resulting in a cumbersome guidance process and low level of intelligence. Summary of the Invention

[0003] The purpose of this invention is to overcome the shortcomings of the prior art. This invention provides a smart guidance method and system for urban services, which improves the intelligence level of service guidance and thus provides users with a more satisfactory user experience.

[0004] To address the aforementioned technical problems, this invention provides a smart guidance method for urban services, the method comprising: Obtain the service request description information input by the user, and preprocess the service request description information to obtain the preprocessed service request description information; A demand semantic graph is constructed based on the preprocessed service demand description information, and the city service knowledge graph is used to identify the target intent information based on the demand semantic graph. Based on the preprocessed service demand description information, a disambiguation mapping table is determined using the city service knowledge graph, and professional term matching is performed based on the disambiguation mapping table to obtain the target professional terms. Based on the target intent information and target professional terms, the preceding item chain is determined, and based on the preceding item chain, logical reasoning of the city service service type is performed to obtain the service dependency graph. Based on the service dependency graph, the city's digital twin service space is used to analyze the guidance path of city services, obtain the target guidance path, and execute the user's city service guidance based on the target guidance path.

[0005] Optionally, the step of preprocessing the service requirement description information to obtain preprocessed service requirement description information includes: The service requirement description information is processed by data integration to obtain the service requirement description information after data integration. Geographically label the service demand description information after data integration and processing to obtain preprocessed service demand description information.

[0006] Optionally, the step of constructing a demand semantic graph based on the preprocessed service demand description information, and using a city service knowledge graph to perform intent recognition based on the demand semantic graph to obtain target intent information, includes: Semantic role labeling is performed based on the preprocessed service requirement description information to obtain semantic role labeling results; Construct a requirement semantic graph based on the semantic role annotation results; Based on the aforementioned demand semantic graph, a candidate item subgraph analysis is performed using the city service knowledge graph to obtain a set of candidate item subgraphs. Based on the candidate item subgraph set, the demand semantic graph is used to filter intents and obtain a sorted list of intent candidates. The confidence level of the candidate intent list is filtered to obtain the confidence level filtering result, and the target intent information is determined based on the confidence level filtering result.

[0007] Optionally, constructing the demand semantic graph based on the semantic role annotation results includes: Based on the semantic role annotation results, a set of semantic triples is determined, and string matching is performed based on the set of semantic triples to obtain the string matching results; A standard entity list is determined based on the string matching results, and standard entity nodes are constructed based on the standard entity list; A requirement semantic graph is constructed based on the standard entity nodes.

[0008] Optionally, the step of determining the disambiguation mapping table based on the preprocessed service demand description information using a city service knowledge graph includes: Keyword extraction is performed based on the preprocessed service requirement description information to obtain a keyword set; The keyword set is labeled with colloquial tags to obtain a keyword set with colloquial tags; Based on the set of keywords annotated with spoken language tags, a disambiguation mapping table is determined using a city service knowledge graph.

[0009] Optionally, the keyword set based on spoken language tags is used to determine a disambiguation mapping table using a city service knowledge graph, including: A limited matching space is determined based on the target intent information; Based on the keyword set annotated with spoken language tags, professional term analysis is performed using the defined matching space to obtain a candidate professional term list; Based on the candidate term list, a mapping process is performed to obtain a mapping table, and the mapping table is then disambiguated to obtain a disambiguated mapping table.

[0010] Optionally, the step of determining the prerequisite item chain based on the target intent information and target professional terms, and performing logical reasoning on the city service service type based on the prerequisite item chain to obtain a service dependency graph, includes: A set of item nodes is constructed based on the target intent information and target professional terms, and an attribute knowledge package is constructed based on the set of item nodes. The preceding event chain is determined based on the event node set and attribute knowledge package, and the parallel event node and triggering event node are determined based on the preceding event chain; Based on the aforementioned preceding item chain, parallel item nodes, and triggering item nodes, logical reasoning is performed on the types of urban service transactions to obtain a service dependency graph.

[0011] Optionally, the step of performing logical reasoning on the city service service type based on the preceding service chain, parallel service nodes, and triggering service nodes to obtain a service dependency graph includes: Construct an initial task dependency graph based on the aforementioned prerequisite task chain, parallel task node, and triggering task node; The initial service dependency graph is pruned to obtain the pruned initial service dependency graph. Based on the pruned initial task dependency graph, the task nodes are adjusted and merged to obtain an intermediate dependency graph. Based on the intermediate dependency graph, logical reasoning is performed on the types of city service applications to obtain the service dependency graph.

[0012] Optionally, the step of analyzing the guidance path for urban services using the urban digital twin service space based on the service dependency graph to obtain the target guidance path includes: Retrieve the city's digital twin service space, and analyze the spatiotemporal accessibility constraints based on the aforementioned service dependency graph using the city's digital twin service space; Based on the spatiotemporal accessibility constraints, a guidance path planning graph is constructed, and a multi-constraint path search is performed based on the guidance path planning graph to obtain a set of candidate guidance paths. A weighted comprehensive evaluation is performed on each candidate guidance path in the candidate guidance path set to obtain the weighted comprehensive evaluation result; Based on the weighted comprehensive evaluation results, the target guidance path is selected from the set of candidate guidance paths.

[0013] In addition, the present invention also provides an intelligent guidance system for urban services, the system comprising: Information preprocessing module: used to acquire service requirement description information input by the user, and preprocess the service requirement description information to obtain preprocessed service requirement description information; Intent recognition module: used to construct a demand semantic graph based on preprocessed service demand description information, and to perform intent recognition based on the demand semantic graph using the city service knowledge graph to obtain target intent information; Specialty term matching module: It is used to determine the disambiguation mapping table based on the preprocessed service demand description information using the city service knowledge graph, and to perform specialty term matching based on the disambiguation mapping table to obtain the target specialty term; Logical reasoning module: used to determine the preceding item chain based on the target intent information and target professional terms, and to perform logical reasoning on the city service service type based on the preceding item chain to obtain the service dependency graph; City Service Guidance Module: This module is used to analyze the guidance path of city services based on the service dependency graph and the city's digital twin service space, obtain the target guidance path, and execute the user's city service guidance based on the target guidance path.

[0014] In this embodiment of the invention, the service request description information input by the user is preprocessed. A request semantic graph is constructed based on the preprocessed service request description information. Intent recognition is then performed using a city service knowledge graph based on the request semantic graph, overcoming the limitations of single text matching and improving the accuracy and robustness of intent recognition. A disambiguation mapping table is determined using the city service knowledge graph based on the preprocessed service request description information. Professional term matching is then performed based on the disambiguation mapping table, providing precise semantic anchors for subsequent logical reasoning and eliminating ambiguities caused by differences in expression. Prerequisite links are determined based on the target intent information and target professional terms. Logical reasoning of city service types is then performed based on these prerequisite links, generating a service dependency graph that clearly presents the complete service chain, allowing users to grasp the overall process without manual review. Based on the service dependency graph, the city's digital twin service space is used to analyze the guidance path for city services, obtaining the target guidance path. The user's city service guidance is then executed based on the target guidance path, avoiding cumbersome city service guidance processes, improving the intelligence of service guidance, and thus providing users with a more satisfactory user experience. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart illustrating the intelligent guidance method for urban services in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structural composition of the intelligent guidance system for urban services in an embodiment of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example 1 Please see Figure 1 , Figure 1 This is a flowchart illustrating the intelligent guidance method for urban services in an embodiment of the present invention. The method includes: S11: Obtain the service requirement description information input by the user, and preprocess the service requirement description information to obtain the preprocessed service requirement description information; In a specific implementation of the present invention, the step of preprocessing the service demand description information to obtain preprocessed service demand description information includes: performing data integration processing on the service demand description information to obtain data integration processed service demand description information; and performing regional tagging on the data integration processed service demand description information to obtain preprocessed service demand description information.

[0019] Specifically, the system acquires service request descriptions from users by simultaneously opening three acquisition channels—voice input, text input, and image capture—through the smart terminal application interface. This allows the system to receive user descriptions of their desired city services, which can be a multimodal input set consisting of voice streams, text strings, and image files.

[0020] The service request description information is processed through data integration. For the speech stream, a streaming automatic speech recognition engine is invoked for real-time transcription, generating a spoken text sequence with timestamps, while preserving sentence pause boundaries and tone features during the transcription process. For image files, a scene classification model is first used to determine whether the image content belongs to the category of certificate reproduction, table screenshot, on-site photo, or handwritten note. Then, based on the classification results, the corresponding optical character recognition model is invoked to extract the text content in the image. At the same time, an object detection model is used to identify service markers such as official seals and QR codes appearing in the image, and the text content and marker information are merged into a structured image description text. The spoken text sequence, image description text, and text input text are then concatenated into a complete original request text stream according to their respective timestamps and user interaction order, with source modality tags added to each segment during the concatenation process. The original request text stream undergoes sentence segmentation and word segmentation, dividing the continuous text into independent sentences. A pre-trained general language model is used to correct colloquial errors in each sentence, replacing typos and non-standard abbreviations with standard written expressions, while retaining the core service-related terms mentioned by the user, resulting in a corrected text stream. Cross-sentence reference resolution is then performed on the corrected text stream, identifying the referents of pronouns in the text and completing the references based on entity nouns and image descriptions. Simultaneously, based on conversational ellipsis rules, omitted subjects and service recipients are explicitly filled in, thus obtaining the integrated service request description information.

[0021] The service requirement description information after data integration and processing is labeled with regional tags. Geographical location fragments are extracted from the service requirement description information after data integration and processing. The extraction results are compared and verified with the user's authorized location information and registered permanent residence administrative division. Ambiguous location descriptions are uniformly replaced with standard four-level administrative division names of province, city, district, and street. Missing administrative division descriptions are automatically completed by default, resulting in text with complete regional tags. At the same time, the text with complete regional tags can be checked for sentence fluency. If unreasonable semantic jumps are detected, completion and re-correction processing is triggered. After the verification is passed, the preprocessed service requirement description information is obtained.

[0022] S12: Construct a demand semantic graph based on the preprocessed service demand description information, and use the city service knowledge graph to perform intent recognition based on the demand semantic graph to obtain target intent information; In a specific implementation of this invention, the step of constructing a demand semantic graph based on preprocessed service demand description information and performing intent recognition using a city service knowledge graph based on the demand semantic graph to obtain target intent information includes: performing semantic role labeling based on preprocessed service demand description information to obtain semantic role labeling results; constructing a demand semantic graph based on the semantic role labeling results; performing candidate item subgraph analysis using a city service knowledge graph based on the demand semantic graph to obtain a candidate item subgraph set; performing intent filtering using the demand semantic graph based on the candidate item subgraph set to obtain an intent candidate ranking list; performing confidence filtering on the intent candidate ranking list to obtain a confidence filtering result, and determining the target intent information based on the confidence filtering result.

[0023] Specifically, semantic role labeling is performed based on the preprocessed service requirement description information. A deep semantic dependency analyzer is used to parse the semantic roles of each sentence in the preprocessed service requirement description information, and the semantic roles of the agent, patient, time constraint, location constraint and manner of the core predicate in each sentence are labeled to obtain the semantic role labeling results.

[0024] Based on the semantic role annotation results, a requirement semantic graph is constructed. The semantic role annotation results can be used to determine the set of semantic triples for string matching; based on the string matching results, a standard entity list is determined to construct standard entity nodes; and based on the standard entity nodes, a requirement semantic graph is constructed.

[0025] Based on the aforementioned demand semantic graph, candidate item subgraph analysis is performed using the city service knowledge graph. The demand semantic graph is used as a query graph, and intent search based on subgraph matching is performed in the service item subgraph of the city service knowledge graph. The action nodes and core recipient nodes in the query graph are used as matching anchors to retrieve all candidate item subgraphs with the same action type and recipient structure, forming a candidate item subgraph set.

[0026] Based on the candidate item subgraph set, the demand semantic graph is used to filter intents. For each candidate item subgraph in the candidate item subgraph set, the graph structure similarity score between it and the demand semantic graph is calculated. The three indicators of node concept overlap, relationship edge consistency and constraint coverage are comprehensively considered. At the same time, the policy validity label of the user's administrative region is introduced to filter out items that have been canceled or suspended. The high-frequency items are weighted and sorted in combination with the user's historical processing records, and the intent candidate ranking list is output.

[0027] The confidence level of the candidate intent ranking list is filtered to obtain the confidence level filtering result. Based on the confidence level filtering result, the target intent information is determined. The item with the highest confidence level and that has passed the regional validity verification is selected from the candidate intent ranking list as the target intent. The target intent information is output. The target intent information may include intent category label, unique code of the corresponding item node in the knowledge graph, standard name of the item, and text description of the basic acceptance conditions of the item.

[0028] Furthermore, the step of constructing a demand semantic graph based on the semantic role annotation results includes: determining a set of semantic triples based on the semantic role annotation results, and performing string matching based on the set of semantic triples to obtain string matching results; determining a standard entity list based on the string matching results, and constructing standard entity nodes based on the standard entity list; and constructing a demand semantic graph based on the standard entity nodes.

[0029] Specifically, a set of semantic triples is determined based on the semantic role annotation results. According to the semantic role annotation results for each sentence, the text fragments corresponding to the noun semantic roles such as agent and patient in the annotation results are extracted as head entities or tail entities. The relation types determined by the core predicate and semantic role types are used as relation words, forming a set of noun-level semantic triples consisting of head entities, relation words, and tail entities. String matching is performed based on the set of semantic triples. The head entity text fragments and tail entity text fragments of each triple in the set of semantic triples are determined as entity mentions to be matched. Each entity mention to be matched is then subjected to fuzzy string matching with the entity node names in the pre-constructed city service knowledge graph to obtain the target string matching results corresponding to each entity mention.

[0030] Based on the string matching results, a standard entity list is determined. The string matching results of each entity mention are fused with the vector similarity calculation results simultaneously calculated for that entity mention, and a comprehensive score is obtained. A set of knowledge graph standard entities is selected as candidates for each entity mention in descending order of score. Simultaneously, for non-standard spoken entity mentions with scores below a threshold, a concept generalization search is performed along pre-built synonym and hypernym / hypernym edges in the knowledge graph. The found hypernym / hyponym nodes are added to the candidate set, ultimately determining a candidate standard entity list composed of knowledge graph standard entities for each entity mention. Standard entity nodes are constructed based on this standard entity list. The best-matching knowledge graph node is selected from the standard entity list, and the corresponding head and tail entity text fragments in the initial semantic triples are directly replaced with this best-matching knowledge graph node, completing the mapping from natural language mentions to knowledge graph standard entity nodes and establishing standard entity nodes.

[0031] Based on the standard entity nodes, a demand semantic graph is constructed. Taking the core action standard entity node obtained by mapping in the user's request as the center, the semantic role relationship such as agent, patient, location, and time is used as directed edges to connect all relevant standard entity nodes, thus constructing a demand semantic graph in which all nodes uniquely correspond to standard concept nodes of the city service knowledge graph.

[0032] S13: Based on the preprocessed service demand description information, a disambiguation mapping table is determined using the city service knowledge graph, and professional term matching is performed based on the disambiguation mapping table to obtain the target professional term; In a specific implementation of this invention, the step of determining a disambiguation mapping table based on preprocessed service demand description information using a city service knowledge graph includes: extracting keywords based on preprocessed service demand description information to obtain a keyword set; labeling the keyword set with colloquial tags to obtain a keyword set with colloquial tags; and determining a disambiguation mapping table based on the keyword set with colloquial tags using a city service knowledge graph.

[0033] Specifically, keyword extraction is performed based on the preprocessed service requirement description information. Shallow syntactic analysis is then conducted on the preprocessed service requirement description information to extract all noun phrases and verb phrases appearing in the text. Stop words and modal particles in the preset stop word list are filtered out to form a user keyword set, i.e., the keyword set is obtained. At the same time, the context word window of the text in which each keyword in the keyword set is located can be retained.

[0034] The keyword set is labeled with colloquial tags. The constructed colloquial segment recognition model is used to detect each keyword in the keyword set one by one to determine whether the keyword belongs to the colloquial terms, abbreviations or informal expressions commonly used by business personnel in the field of business. The identified colloquial keywords are labeled with colloquial tags and their character positions in the original text are recorded to obtain the keyword set after colloquial tagging.

[0035] Based on the keyword set annotated with spoken language tags, a disambiguation mapping table is determined using a city service knowledge graph, and a limited matching space can be determined through target intent information. Based on the keyword set annotated with spoken language tags, professional term analysis is performed using the limited matching space to obtain a list of candidate professional terms. Based on the list of candidate professional terms, mapping processing is performed to obtain a mapping table, and disambiguation processing is performed on the mapping table to obtain a disambiguation mapping table.

[0036] Based on the disambiguation mapping table, professional term matching is performed. For each candidate professional term retained in the disambiguation mapping table, a secondary sorting is performed according to the association weight between it and the target intent item node in the knowledge graph. Professional terms that frequently appear as professional requirements in the application materials, approval conditions, or procedures are given priority. A predetermined number of professional terms that rank at the top after sorting are determined as the target professional term list. Each target professional term in the list contains a standard term name, associated item code, and responsible department information, thus obtaining the target professional term.

[0037] Furthermore, the step of determining a disambiguation mapping table using a city service knowledge graph based on the keyword set annotated with spoken language tags includes: determining a limited matching space based on the target intent information; performing professional term analysis using the limited matching space based on the keyword set annotated with spoken language tags to obtain a candidate professional term list; performing mapping processing based on the candidate professional term list to obtain a mapping table; and performing disambiguation processing on the mapping table to obtain a disambiguation mapping table.

[0038] Specifically, a limited matching space is determined based on the target intent information. According to the intent category label and the unique code of the item node contained in the target intent information, the professional term subdomain associated with the intent is located and activated in the city service knowledge graph. The word vector model and term mapping table exclusive to the subdomain are loaded to form a limited matching space.

[0039] Based on the keyword set annotated with spoken language tags, professional term analysis is performed using the defined matching space. Each keyword in the keyword set annotated with spoken language tags is matched one by one with the standard business term node name of the subdomain in the knowledge graph within the defined matching space. Vector encoding and semantic similarity calculation are performed to obtain a term similarity score list for each keyword, and a candidate professional term list is generated for each keyword.

[0040] Based on the candidate terminology list, mapping processing is performed. For keywords tagged with colloquialisms, the pre-built colloquial-standard name relationship edges in the city service knowledge graph are queried first. If the relationship edge directly connects the colloquial keyword to a unique standard terminology node within the limited matching space, then the standard terminology node is extracted as another candidate terminology for that keyword, and its candidate priority is set higher than that of the candidate terms in the candidate terminology list. The candidate terminology list and the other candidate terminology list are merged to establish a mapping table from each user keyword to several candidate standard terms. Each candidate terminology entry in the table carries its node identifier, standard terminology name, and responsible department information in the knowledge graph. The mapping table is then subjected to disambiguation processing. Co-occurrence disambiguation processing is performed on the mapping table. Combining the overall context window of the user's original requirement description, the correlation strength between each pair of candidate terms for different keywords in the mapping table in the knowledge graph is calculated. Candidate terms that semantically conflict with co-occurring terms or whose correlation strength is lower than a preset threshold are eliminated using contextual semantic consistency constraints to obtain a disambiguation mapping table.

[0041] S14: Determine the prerequisite items link based on the target intent information and target professional terms, and perform logical reasoning on the city service service type based on the prerequisite items link to obtain the service dependency graph; In a specific implementation of this invention, the step of determining the prerequisite item chain based on the target intent information and target professional terms, and performing logical reasoning on the city service service type based on the prerequisite item chain to obtain a service dependency graph, includes: constructing a set of item nodes based on the target intent information and target professional terms; constructing an attribute knowledge package based on the set of item nodes; determining the prerequisite item chain based on the set of item nodes and the attribute knowledge package; determining parallel item nodes and triggering item nodes based on the prerequisite item chain; and performing logical reasoning on the city service service type based on the prerequisite item chain, parallel item nodes, and triggering item nodes to obtain a service dependency graph.

[0042] Specifically, a set of item nodes is constructed based on the target intent information and target professional terms. Using the unique code of the item node in the target intent information as an index, the corresponding target intent item node is extracted from the city service knowledge graph and set as the main root node for the service logic reasoning. At the same time, the list of target professional terms is traversed, and the service item nodes directly associated with each target professional term in the knowledge graph are queried. All the queried service item nodes are set as additional nodes, and the main root node and additional nodes are merged to form a set of item nodes.

[0043] Based on the set of item nodes, an attribute knowledge package is constructed. The service process model attributes of each item node in the set of item nodes are extracted from the city service knowledge graph. The attribute content includes the legal processing sequence of the item, the list of materials required for each step, the processing form requirements, and the dependency relationship edge type between the item node and other item nodes in the knowledge graph, thus forming the attribute knowledge package.

[0044] Based on the aforementioned set of item nodes and attribute knowledge package, the preceding item chain is determined. The set of item nodes and attribute knowledge package are input into the service logic reasoning engine. The reasoning engine performs a depth-first multi-hop path search along the preceding item relationship edges defined in the knowledge graph, starting from the root node. It traces upwards layer by layer from the root node, tracing all preceding item nodes that must be processed first. Each preceding item is used as a new starting point to continue tracing upwards until an initial node with no preceding items is reached, thus obtaining the preceding item chain. Based on the preceding item chain, parallel item nodes and triggering item nodes are determined. While performing preceding item tracing, the reasoning engine expands its search outwards along the parallel item relationship edges defined in the knowledge graph, identifying parallel item nodes that belong to the same business stage as the root node or any preceding item node and are allowed to be applied for simultaneously. The identified parallel item nodes are assigned to their corresponding parallel groups and labeled with group identifiers. Simultaneously, along the conditional triggering item relationship edges, it determines whether the user meets the trigger threshold under the current condition set, adding triggering item nodes that meet the trigger threshold to the node set.

[0045] Logical reasoning of city service types is performed based on the aforementioned prerequisite item links, parallel item nodes, and triggering item nodes. An initial service dependency graph is constructed based on the prerequisite item links, parallel item nodes, and triggering item nodes. The initial service dependency graph is pruned to adjust and merge item nodes to obtain an intermediate dependency graph. Logical reasoning of city service types is performed based on the intermediate dependency graph to obtain a service dependency relationship graph.

[0046] Furthermore, the step of performing logical reasoning on the city service service type based on the preceding item chain, parallel item nodes, and triggering item nodes to obtain a service dependency graph includes: constructing an initial service dependency graph based on the preceding item chain, parallel item nodes, and triggering item nodes; pruning the initial service dependency graph to obtain a pruned initial service dependency graph; adjusting and merging item nodes based on the pruned initial service dependency graph to obtain an intermediate dependency graph; and performing logical reasoning on the city service service type based on the intermediate dependency graph to obtain a service dependency graph.

[0047] Specifically, an initial service dependency graph is constructed based on the aforementioned prerequisite task chain, parallel task nodes, and triggering task nodes. All task nodes and parallel task nodes in the prerequisite task chain are connected to the triggering task nodes in a graph structure according to the serial dependency relationship and parallel relationship extracted from the knowledge graph. The serial dependency is represented by directed edges to indicate the order in which tasks must be performed first, and the parallel relationship is represented by undirected edges to indicate the parallel relationship that can be recommended synchronously, forming an initial service dependency graph with task nodes as units and relationship edges as connections.

[0048] The initial service dependency graph is pruned. The user's electronic certificate database and personal information authorization profile system are queried to obtain the list of electronic certificates currently held by the user, the list of approved documents, and the historical service results records of completed procedures. The prerequisite material requirements marked for each item node in the initial service dependency graph are compared item by item with the list of certificates held by the user. Item nodes for which all prerequisite materials have been met are marked as met and can be skipped. The node is then pruned from the dependency graph to obtain the pruned initial service dependency graph.

[0049] Based on the pruned initial service dependency graph, the item nodes are adjusted and merged. The current effective administrative approval improvement rule library of the user's administrative region is retrieved. Each item node retained in the pruned initial service dependency graph is matched with the list of notification and commitment system items, the list of certificate separation items, and the list of multi-certificate integration items in the improvement rule library. The matching item nodes are marked with the corresponding improvement strategy. According to the multi-certificate integration rule, multiple integrable item nodes are merged into one comprehensive acceptance node. The dependency relationship after merging is adjusted to obtain the intermediate dependency graph.

[0050] Based on the intermediate dependency graph, logical reasoning is performed on the types of urban service procedures. All item nodes in the intermediate dependency graph are topologically sorted according to the order indicated by the serial dependency edges. Parallel items with the same topological order are grouped into the same processing stage layer. Each item node is supplemented with annotations of the opinion processing channel, responsible management department, statutory commitment time limit, actual time limit after the application of the notification and commitment system, and a complete list of materials after summarization. The final directed acyclic service dependency graph is then output, thus obtaining the service dependency graph.

[0051] S15: Based on the service dependency graph, analyze the guidance path of urban services using the urban digital twin service space to obtain the target guidance path, and execute the user's urban service guidance based on the target guidance path.

[0052] In a specific implementation of this invention, the step of analyzing the guidance path for urban services using the urban digital twin service space based on the service dependency graph to obtain the target guidance path includes: retrieving the urban digital twin service space; analyzing spatiotemporal accessibility constraints based on the service dependency graph using the urban digital twin service space; constructing a guidance path planning graph based on the spatiotemporal accessibility constraints; performing a multi-constraint path search based on the guidance path planning graph to obtain a set of candidate guidance paths; performing a weighted comprehensive evaluation on each candidate guidance path in the candidate guidance path set to obtain a weighted comprehensive evaluation result; and selecting the target guidance path from the set of candidate guidance paths based on the weighted comprehensive evaluation result.

[0053] Specifically, the city's digital twin service space is retrieved, and spatiotemporal accessibility constraints are analyzed based on the service dependency graph. The pre-constructed city digital twin service space is retrieved, and each item node in the service dependency graph is matched and bound to the list of acceptable items of each service point in the digital twin space to generate a list of service resources that can be processed for each item. At the same time, the user's real-time location and travel preferences are obtained, the estimated travel time and arrival time window of the user to each physical service point are calculated, and the required terminal environment requirements are marked for the online service entry. The spatiotemporal accessibility constraints that include virtual and physical service resources and their spatiotemporal accessibility attributes are comprehensively formed.

[0054] Based on the aforementioned spatiotemporal accessibility constraints, a guidance path planning graph is constructed. The action of completing the processing of each item at a specific service point within a specific time period is used as the basic decision unit. The dependency order between items is used as serial connection edges, and the spatial transfer path across service points is used as transfer connection edges, thus constructing a guidance path planning graph with spatiotemporal cost attributes and channel cost attributes. Based on this guidance path planning graph, a multi-constraint path search is performed. Time windows are allocated to each stage on the guidance path planning graph according to the topological stage order of the service dependency graph. Service point aggregation optimization is performed on parallel items in the same stage to reduce the number of transfers. The multi-constraint path search algorithm generates several candidate guidance paths carrying complete step sequences and time scales, forming a candidate guidance path set.

[0055] A weighted comprehensive evaluation is performed on each candidate guidance path in the candidate guidance path set. For each candidate guidance path in the candidate guidance path set, evaluation indicators such as total time spent in the whole process, total walking distance of the user, number of transfers across service points, and matching degree of urgency of each matter are extracted. The comprehensive value of each candidate guidance path is calculated by weighting and summing according to the weight coefficients of each indicator in the user's preset objective function, and a weighted comprehensive evaluation result containing the comprehensive score of each candidate path is formed.

[0056] Based on the weighted comprehensive evaluation results, a target guidance path is selected from the candidate guidance path set. The comprehensive value of each candidate guidance path in the weighted comprehensive evaluation results is ranked and compared. The path with the smallest comprehensive value, i.e. the best comprehensive value, is selected as the target guidance path from the candidate guidance path set. The target guidance path may include a complete sequence of processing steps, the online or offline channel type corresponding to each step, the name and window number of the service point or the online entry link, the time of opinion arrival, and the expected completion time of each step.

[0057] Based on the target guidance path, the system guides users to access city services. The target guidance path is broken down into a set of step-by-step guidance instructions and pushed to the user's smart terminal. These instructions are presented interactively in the form of visual service location markers and a time-series task list. During the user's execution, the system continuously monitors the real-time status of each service point through a digital twin service space. When an anomaly is detected at a service point, the system triggers dynamic replanning of the local path and pushes an update instruction to the user in real time. In addition, before each step is executed, the system automatically retrieves the corresponding electronic form for intelligent pre-filling and assists in completing the pre-review of materials and appointment queuing. The system continues to track until all items in the service dependency graph are completed.

[0058] In this embodiment of the invention, the service request description information input by the user is preprocessed. A request semantic graph is constructed based on the preprocessed service request description information. Intent recognition is then performed using a city service knowledge graph based on the request semantic graph, overcoming the limitations of single text matching and improving the accuracy and robustness of intent recognition. A disambiguation mapping table is determined using the city service knowledge graph based on the preprocessed service request description information. Professional term matching is then performed based on the disambiguation mapping table, providing precise semantic anchors for subsequent logical reasoning and eliminating ambiguities caused by differences in expression. Prerequisite links are determined based on the target intent information and target professional terms. Logical reasoning of city service types is then performed based on these prerequisite links, generating a service dependency graph that clearly presents the complete service chain, allowing users to grasp the overall process without manual review. Based on the service dependency graph, the city's digital twin service space is used to analyze the guidance path for city services, obtaining the target guidance path. The user's city service guidance is then executed based on the target guidance path, avoiding cumbersome city service guidance processes, improving the intelligence of service guidance, and thus providing users with a more satisfactory user experience.

[0059] Example 2 Please see Figure 2 , Figure 2 This is a schematic diagram of the structural composition of the intelligent urban service guidance system in an embodiment of the present invention. The system includes: Information preprocessing module 21: used to obtain service requirement description information input by the user, and preprocess the service requirement description information to obtain preprocessed service requirement description information; Intent recognition module 22: used to construct a demand semantic graph based on preprocessed service demand description information, and to perform intent recognition based on the demand semantic graph using a city service knowledge graph to obtain target intent information; Specialty term matching module 23: It is used to determine a disambiguation mapping table based on the preprocessed service demand description information using the city service knowledge graph, and to perform specialty term matching based on the disambiguation mapping table to obtain the target specialty term; Logical reasoning module 24: used to determine the preceding item chain based on the target intent information and target professional terms, and to perform logical reasoning on the city service service type based on the preceding item chain to obtain the service dependency graph; City Service Guidance Module 25: This module is used to analyze the guidance path of city services based on the service dependency graph and the city's digital twin service space, obtain the target guidance path, and execute the user's city service guidance based on the target guidance path.

[0060] In the specific implementation of this invention, the specific implementation methods of the system items can be referred to the implementation methods of the above-mentioned method items, and will not be repeated here.

[0061] In this embodiment of the invention, the service request description information input by the user is preprocessed. A request semantic graph is constructed based on the preprocessed service request description information. Intent recognition is then performed using a city service knowledge graph based on the request semantic graph, overcoming the limitations of single text matching and improving the accuracy and robustness of intent recognition. A disambiguation mapping table is determined using the city service knowledge graph based on the preprocessed service request description information. Professional term matching is then performed based on the disambiguation mapping table, providing precise semantic anchors for subsequent logical reasoning and eliminating ambiguities caused by differences in expression. Prerequisite links are determined based on the target intent information and target professional terms. Logical reasoning of city service types is then performed based on these prerequisite links, generating a service dependency graph that clearly presents the complete service chain, allowing users to grasp the overall process without manual review. Based on the service dependency graph, the city's digital twin service space is used to analyze the guidance path for city services, obtaining the target guidance path. The user's city service guidance is then executed based on the target guidance path, avoiding cumbersome city service guidance processes, improving the intelligence of service guidance, and thus providing users with a more satisfactory user experience.

[0062] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, which may include: read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, etc.

[0063] Furthermore, the above provides a detailed description of the intelligent guidance method and system for urban services provided by the embodiments of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A smart guidance method for urban services, characterized in that, The method includes: Obtain the service request description information input by the user, and preprocess the service request description information to obtain the preprocessed service request description information; A demand semantic graph is constructed based on the preprocessed service demand description information, and the city service knowledge graph is used to identify the target intent information based on the demand semantic graph. Based on the preprocessed service demand description information, a disambiguation mapping table is determined using the city service knowledge graph, and professional term matching is performed based on the disambiguation mapping table to obtain the target professional terms. Based on the target intent information and target professional terms, the preceding item links are determined, and logical reasoning of the city service types is performed based on the preceding item links to obtain the service dependency graph. Based on the service dependency graph, the city's digital twin service space is used to analyze the guidance path of city services, obtain the target guidance path, and execute the user's city service guidance based on the target guidance path.

2. The intelligent guidance method for urban services according to claim 1, characterized in that, The step of preprocessing the service requirement description information to obtain preprocessed service requirement description information includes: The service requirement description information is processed by data integration to obtain the service requirement description information after data integration. Geographically label the service demand description information after data integration and processing to obtain preprocessed service demand description information.

3. The intelligent guidance method for urban services according to claim 1, characterized in that, The process involves constructing a demand semantic graph based on preprocessed service demand description information, and then using a city service knowledge graph to perform intent recognition based on the demand semantic graph to obtain target intent information, including: Semantic role labeling is performed based on the preprocessed service requirement description information to obtain semantic role labeling results; Construct a requirement semantic graph based on the semantic role annotation results; Based on the aforementioned demand semantic graph, a candidate item subgraph analysis is performed using the city service knowledge graph to obtain a set of candidate item subgraphs. Based on the candidate item subgraph set, the demand semantic graph is used to filter intents and obtain a sorted list of intent candidates. The confidence level of the candidate intent list is filtered to obtain the confidence level filtering result, and the target intent information is determined based on the confidence level filtering result.

4. The intelligent guidance method for urban services according to claim 3, characterized in that, The construction of the demand semantic graph based on the semantic role annotation results includes: Based on the semantic role annotation results, a set of semantic triples is determined, and string matching is performed based on the set of semantic triples to obtain the string matching results; A standard entity list is determined based on the string matching results, and standard entity nodes are constructed based on the standard entity list; A requirement semantic graph is constructed based on the standard entity nodes.

5. The intelligent guidance method for urban services according to claim 1, characterized in that, The process of determining a disambiguation mapping table based on preprocessed service demand description information using a city service knowledge graph includes: Keyword extraction is performed based on the preprocessed service requirement description information to obtain a keyword set; The keyword set is labeled with colloquial tags to obtain a keyword set with colloquial tags; Based on the set of keywords annotated with spoken language tags, a disambiguation mapping table is determined using a city service knowledge graph.

6. The intelligent guidance method for urban services according to claim 5, characterized in that, The keyword set annotated with spoken language tags is used to determine a disambiguation mapping table using a city service knowledge graph, including: A limited matching space is determined based on the target intent information; Based on the keyword set annotated with spoken language tags, professional term analysis is performed using the defined matching space to obtain a candidate professional term list; Based on the candidate term list, a mapping process is performed to obtain a mapping table, and the mapping table is then disambiguated to obtain a disambiguated mapping table.

7. The intelligent guidance method for urban services according to claim 1, characterized in that, The process of determining the prerequisite steps based on the target intent information and target professional terms, and performing logical reasoning on the types of city service requests based on the prerequisite steps to obtain a service dependency graph, includes: A set of item nodes is constructed based on the target intent information and target professional terms, and an attribute knowledge package is constructed based on the set of item nodes. The preceding event chain is determined based on the event node set and attribute knowledge package, and the parallel event node and triggering event node are determined based on the preceding event chain; Based on the aforementioned preceding item chain, parallel item nodes, and triggering item nodes, logical reasoning is performed on the types of urban service transactions to obtain a service dependency graph.

8. The intelligent guidance method for urban services according to claim 7, characterized in that, The logical reasoning of city service types based on the preceding event chain, parallel event nodes, and triggering event nodes, to obtain the service dependency graph, includes: Construct an initial task dependency graph based on the aforementioned prerequisite task chain, parallel task node, and triggering task node; The initial service dependency graph is pruned to obtain the pruned initial service dependency graph. Based on the pruned initial task dependency graph, the task nodes are adjusted and merged to obtain an intermediate dependency graph. Based on the intermediate dependency graph, logical reasoning is performed on the types of city service applications to obtain the service dependency graph.

9. The intelligent guidance method for urban services according to claim 1, characterized in that, The step of analyzing the guidance path for urban services based on the service dependency graph using the urban digital twin service space to obtain the target guidance path includes: Retrieve the city's digital twin service space, and analyze the spatiotemporal accessibility constraints based on the aforementioned service dependency graph using the city's digital twin service space; Based on the spatiotemporal accessibility constraints, a guidance path planning graph is constructed, and a multi-constraint path search is performed based on the guidance path planning graph to obtain a set of candidate guidance paths. A weighted comprehensive evaluation is performed on each candidate guidance path in the candidate guidance path set to obtain the weighted comprehensive evaluation result; Based on the weighted comprehensive evaluation results, the target guidance path is selected from the set of candidate guidance paths.

10. A smart guidance system for urban services, employing the smart guidance method for urban services as described in any one of claims 1-9, characterized in that, The system includes: Information preprocessing module: used to acquire service requirement description information input by the user, and preprocess the service requirement description information to obtain preprocessed service requirement description information; Intent recognition module: used to construct a demand semantic graph based on preprocessed service demand description information, and to perform intent recognition based on the demand semantic graph using the city service knowledge graph to obtain target intent information; Specialty term matching module: It is used to determine the disambiguation mapping table based on the preprocessed service demand description information using the city service knowledge graph, and to perform specialty term matching based on the disambiguation mapping table to obtain the target specialty term; Logical reasoning module: used to determine the preceding item chain based on the target intent information and target professional terms, and to perform logical reasoning on the city service service type based on the preceding item chain to obtain the service dependency graph; City Service Guidance Module: This module is used to analyze the guidance path of city services based on the service dependency graph and the city's digital twin service space, obtain the target guidance path, and execute the user's city service guidance based on the target guidance path.

Images