Artificial intelligence-based table generation method, system and equipment and storage medium

Abstract

The invention relates to the technical field of data processing, and particularly provides an artificial intelligence-based table generation method, system and device and a storage medium, and the method comprises the following steps: obtaining text data, and extracting an entity from the text data; determining attributes of the entity through syntactic context analysis, and caching corresponding attribute values; matching a pre-configured user attention item with the attribute, determining a table column according to the successfully matched attribute, and determining a table row according to the number of entities in the adjacent matrix; and generating a table based on the table columns and the table rows, and filling the attribute values of the entities to corresponding positions of the table. The table generated by the method is complete in content and high in structural adaptability, and can effectively meet the requirements for multi-dimensional and customized presentation of information in different scenes.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and particularly relates to a method, system, device and storage medium for generating tables based on artificial intelligence. Background Art

[0002] In the prior art, automated form filling is an important means to improve data processing efficiency. The commonly used method currently usually relies on predefined structured templates. Its core process is: first, identify and extract preset target field information from source data (such as text, database records, etc.); subsequently, directly fill the extracted target field values into the corresponding positions of a pre-configured table template with a fixed format and table items (rows / columns).

[0003] However, this method has significant limitations. First of all, its function is limited to filling simple and discrete field values, such as single information points like name, date, number, etc. For more complex attribute information contained in the data, such as event descriptions, relationships between entities, dynamic states, etc., which have hierarchical relationships or require context understanding to determine, the existing methods are often difficult to effectively identify and extract.

[0004] Secondly, and most critically, its defect lies in its staticness and rigidity. The structure of the table (including table item content, column headers, number of rows, etc.) is completely determined by the pre-configured template, and cannot be dynamically adjusted according to the actual data content extracted, the size of the data volume, or the user's specific and temporary concerns. For example, when the source data contains additional attributes not defined in the template, this valuable information will be ignored; and when the user needs to focus on different data dimensions, the template must be redesigned and configured.

[0005] This form filling method lacking context understanding and dynamic adaptation ability results in incomplete and inflexible table information, and it is difficult to meet the requirements for multi-dimensional and customized presentation of information in complex and changeable scenarios (such as government report generation, financial data analysis, medical record collation, etc.). Therefore, there is an urgent need for a solution that can more intelligently understand data semantics and dynamically construct table structures to adapt to different data and user requirements. Summary of the Invention

[0006] In view of the above deficiencies of the prior art, the present invention provides a method, system, device and storage medium for generating tables based on artificial intelligence to solve the above technical problems.

[0007] In a first aspect, the present invention provides a method for generating a table based on artificial intelligence, including: Obtaining text data and extracting entities from the text data; Determining the attributes of the entities through syntactic context analysis and caching the corresponding attribute values; Match the pre-configured user attention items with the attributes, determine the table columns according to the successfully matched attributes, and determine the table rows according to the number of entities in the adjacency matrix; Generate a table based on the table columns and rows, and fill the attribute values of the entities into the corresponding positions of the table.

[0008] In an alternative embodiment, obtaining text data and extracting entities from the text data includes: Obtain the text data input by the front end; Use named entity recognition technology to identify multiple entities from the text data; Perform clustering processing on the multiple entities, and merge the entities with similarity exceeding a set threshold into the same category of entities; Generate an entity list according to the merged entities.

[0009] In an alternative embodiment, merging the entities with similarity exceeding a set threshold into the same category of entities includes: Assign a unified concept to the entity category, and replace the entity names belonging to the same entity category with the unified concept.

[0010] In an alternative embodiment, determining the attributes of the entities through syntactic context analysis and caching the corresponding attribute values includes: Construct a syntactic dependency tree for the text data; Extract subject-verb-object triples from the syntactic dependency tree; Extract the action pairs containing the entities from the subject-verb-object triples, and use the BiLSTM model to predict the attributes of the entities based on the action pairs; Locate the position of the attribute in the text data, and identify the corresponding attribute value based on the position.

[0011] In an alternative embodiment, the method further includes: Construct a basic adjacency matrix according to the number of entities and the number of attributes; Calculate the co-occurrence times of the entities and attributes in the text data and the semantic similarity between the entities and attributes respectively; Obtain the edge weights of the entities and attributes by calculating the weighted sum of the co-occurrence times and the semantic similarity; Fill the basic adjacency matrix according to the edge weights of the entities and attributes.

[0012] In an alternative embodiment, the method further includes: Obtain the keywords configured by the user, and store the keywords as explicit attention items; Obtain the user operation log, and the user operation log records the operation behaviors of the user on the historically generated tables. Determine the implicit attention items and the attention degree of each table item according to the operation behavior; The attention degree is used to determine the front and back positions of the table items corresponding to the determined attributes in the table.

[0013] In an optional embodiment, matching the pre-configured user attention items with the attributes in the adjacency matrix includes: Calculate the semantic similarity between the user attention items and the attributes respectively, and mark the attributes with the semantic similarity reaching the set similarity threshold as display attributes; the semantic similarity includes explicit similarity and implicit similarity.

[0014] In a second aspect, the present invention provides an artificial intelligence-based table generation system, including: A first recognition module, configured to obtain text data and extract entities from the text data; A second recognition module, configured to determine the attributes of the entities through syntactic context analysis and cache the corresponding attribute values; A relationship construction module, configured to construct an adjacency matrix of entities and attributes; A parameter determination module, configured to match the pre-configured user attention items with the attributes, determine the table columns according to the successfully matched attributes, and determine the table rows according to the number of entities in the adjacency matrix; A table generation module, configured to generate a table based on the table columns and table rows, and fill the attribute values of the entities into the corresponding positions of the table.

[0015] In a third aspect, a device is provided, including: A memory, configured to store an artificial intelligence-based table generation program; A processor, configured to implement the steps of the artificial intelligence-based table generation method provided in the first aspect when executing the artificial intelligence-based table generation program.

[0016] In a fourth aspect, a computer-readable storage medium is provided, on which an artificial intelligence-based table generation program is stored. When the artificial intelligence-based table generation program is executed by a processor, the steps of the artificial intelligence-based table generation method provided in the first aspect are implemented.

[0017] The beneficial effects of the present invention are as follows. The method, system, device and storage medium for generating tables based on artificial intelligence provided by the present invention can, through syntactic context analysis, not only identify simple entities, but also accurately analyze their complex attributes (such as status, relationships), significantly improving the ability to capture deep semantic information in unstructured text. Based on the intelligent matching of user attention items and extracted attributes, the table column headers and the number of rows are automatically determined, completely getting rid of the limitations of predefined templates, and being able to dynamically generate an optimized table structure according to the actual data content and user needs. The tables generated by the present invention have complete content and strong structural adaptability, can effectively meet the requirements for multi-dimensional and customized presentation of information in different scenarios (such as report generation, data analysis), and greatly improve the intelligence level and practicality of table generation. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0019] Figure 1 is a schematic flowchart of the method according to an embodiment of the present invention.

[0020] Figure 2 is a schematic block diagram of the system according to an embodiment of the present invention.

[0021] Figure 3 is a schematic structural diagram of a device provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] In order to enable those skilled in the art of the present technology to better understand the technical solutions in the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0023] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field of the present invention. The terms used in the description of the present invention in this specification are only for the purpose of describing specific embodiments, and are not intended to limit the present invention.

[0024] The method for generating tables based on artificial intelligence provided by the embodiments of the present invention is executed by a computer device. Correspondingly, the system for generating tables based on artificial intelligence runs in the computer device.

[0025] Figure 1 is a schematic flowchart of the method according to an embodiment of the present invention. Among them, Figure 1 The execution subject can be an artificial intelligence-based form generation system. According to different requirements, the order of steps in this flowchart can be changed, and some can be omitted.

[0026] As Figure 1 shown, the method includes: S1. Obtain text data and extract entities from the text data; S2. Determine the attributes of the entities through syntactic context analysis and cache the corresponding attribute values; S3. Match the pre-configured user concern items with the attributes, determine the form columns according to the successfully matched attributes, and determine the form rows according to the number of entities in the adjacency matrix; S4. Generate a form based on the form columns and form rows, and fill the attribute values of the entities into the corresponding positions of the form.

[0027] In an embodiment of the present invention, based on step S1, a possible embodiment will be given below to non-restrictively elaborate on its specific implementation.

[0028] S101. Obtain the text data input from the front end.

[0029] Adopt an event-driven asynchronous data capture mechanism to achieve real-time text stream collection by listening to the HTML5 Input Event interface. Use the throttle technology to limit the data transmission frequency to 50 ms / time to balance the user experience and the backend processing pressure. Use the Web Socket protocol to establish a two-way communication channel to achieve segmented streaming transmission of text data, effectively reducing the transmission delay when a large amount of data is input. The AES-256 algorithm is used for end-to-end encryption during the data transmission process to ensure the security of user privacy data.

[0030] S102. Use named entity recognition technology to identify multiple entities from the text data.

[0031] A hybrid entity recognition model was constructed, using the BERT-large pre-trained language model for word vector representation at the bottom layer and introducing a multi-head attention mechanism to capture long-range dependencies in text. During the NER tag decoding phase, a joint decoding framework was constructed by integrating the Conditional Random Field (CRF) and Bidirectional Long Short-Term Memory (BiLSTM) network to effectively address entity boundary ambiguity. For entity recognition in low-resource domains, a meta-learning-based parameter initialization method was designed, which, through 5-shot learning, can achieve the same training results as traditional methods using only 500 samples. The system supports custom entity type extensions and enables domain knowledge injection through an ontology mapping mechanism.

[0032] S103. Clustering is performed on the plurality of entities, and entities whose similarity exceeds a set threshold are grouped into entities of the same type; a unified concept is assigned to the entity category, and entity names belonging to the same entity category are replaced with the unified concept.

[0033] The entity clustering algorithm based on semantic graph embedding first constructs an entity co-occurrence network and uses the Node2Vec algorithm to generate a low-dimensional vector representation of the entity. The clustering process uses the density peak clustering (DPC) algorithm, and the cluster center and the number of clusters are automatically determined by calculating the decision diagram. The similarity calculation integrates the cosine similarity of word vectors, the similarity of hyponymy, and the similarity of distribution to construct a multidimensional similarity space. The concept merging stage adopts an ontology-driven concept mapping mechanism, constructs a concept hierarchy system through WordNet and DBpedia, and uses the OWL semantic reasoning engine to realize the mapping conversion from entity names to unified concepts. The system supports dynamic threshold adjustment and automatically determines the optimal similarity threshold through information entropy calculation.

[0034] S104. Generate an entity list based on the merged entities.

[0035] A multi-dimensional entity list generation algorithm was designed. At the data structure level, a prefix tree (Trie) was used for efficient entity retrieval, supporting entity prefix matching with O(k) time complexity. The list display employed a hierarchical visualization strategy, with multi-dimensional sorting based on entity category, confidence, and frequency of occurrence. To enhance the human-computer interaction experience, an entity importance ranking algorithm based on an attention mechanism was implemented, automatically placing key entities at the top of the display. For data persistence, a Neo4j graph database was used to store semantic relationships between entities, supporting the subsequent construction of a knowledge graph.

[0036] In an embodiment of the present invention, based on step S2, a possible embodiment will be given below to illustrate its specific implementation scheme in a non-limiting manner.

[0037] S201. Construct a syntactic dependency tree for the text data.

[0038] A syntactic analysis model based on a neural network is used for text data processing. This model takes a pre-trained Transformer encoder (such as RoBERTa-large) as the backbone network, and captures the long-distance dependencies between words in the text through the multi-head attention mechanism. During the training process, a dependency relationship prediction module based on Biaffine attention is introduced. The context representation vectors of each word are used as inputs, and the dependency relationship score matrix is calculated through a two-layer fully connected network, and then the dependency relationship labels between words are predicted. To improve the adaptability of the model to different language structures, multi-language pre-training and domain adaptation fine-tuning strategies are adopted. Based on the general domain pre-trained model, specific domain corpus is used for fine-tuning. At the same time, by introducing a syntactic constraint loss function, the model's learning of syntactic rules is enhanced to ensure that the generated syntactic dependency tree conforms to linguistic norms.

[0039] S202. Extract the subject-verb-object triples from the syntactic dependency tree.

[0040] Based on the constructed syntactic dependency tree, a triple extraction algorithm combining rules and statistics is designed. First, the subject-verb-object extraction rule templates are formulated according to linguistic definitions. For example, rules are constructed with dependency relationship labels such as "nsubj (nominal subject)", "dobj (direct object)", "root (sentence core verb)", etc. For each sentence, traverse the syntactic dependency tree, match the node combinations that meet the conditions according to the rule templates, and extract potential subject-verb-object triples. To handle complex sentence patterns and semantic ambiguity, a statistical model is introduced for auxiliary verification. By counting the common subject-verb-object structure patterns and their co-occurrence frequencies in a large-scale corpus, a probability model is constructed. For the candidate triples extracted by the rules, calculate their confidence scores under the statistical model, set a threshold to filter out low-confidence results, and finally obtain an accurate set of subject-verb-object triples.

[0041] S203. Extract the action pairs containing entities from the subject-verb-object triples, and use the BiLSTM model to predict the attributes of the entities based on the action pairs.

[0042] From the extracted subject-predicate-object triples, filter out the action pairs (i.e., subject-predicate or verb-object structures) that contain the recognized entities. Take the lexical sequences in these action pairs as input for word vector representation. Use a pre-trained word vector model (such as GloVe) to map each word to a low-dimensional dense vector, and incorporate the position information of the word in the sentence into the vector representation through positional encoding. Input the processed vector sequence into a bidirectional long short-term memory network (BiLSTM). The BiLSTM captures the contextual semantic information of the words in the action pair through forward and backward propagation, and outputs a hidden state vector at each time step. At the output layer, use a fully connected layer to map the final hidden state vector to a probability distribution of attribute categories, calculate the probability of each attribute category through the Softmax function, and select the category with the highest probability as the predicted entity attribute. To alleviate the problem of model overfitting, introduce the Dropout regularization technique into the network and adopt the EarlyStopping strategy to optimize the model training process.

[0043] S204. Locate the position of the attribute in the text data and identify the corresponding attribute value based on the position.

[0044] After predicting the entity attribute, perform attribute value location based on the syntactic dependency tree and the original text position information. First, according to the predicted attribute category, find the syntactic structure path related to the attribute in the syntactic dependency tree. For example, for descriptive attributes, usually look for them in the structure modified by adjectives or adverbs; for relational attributes, locate them based on relevant prepositional phrases or clause structures. Then, combine the character offset information of the text to determine the start and end positions of the attribute value in the original text data. To improve the location accuracy, introduce a conditional random field (CRF) model to optimize the preliminary location results. Take features such as the word vectors, part-of-speech tags, and syntactic dependency relation tags of the text as the input of the CRF, and predict the most likely boundary label sequence of the attribute value by learning the transition probabilities and feature functions between the labels, so as to accurately identify the attribute value.

[0045] In an embodiment of the present invention, the relationship between the entity and the attribute can be stored as an adjacency matrix.

[0046] 1. Construct a basic adjacency matrix according to the number of entities and the number of attributes.

[0047] Let the entity set be E = {e1, e2, …, e m}, and the attribute set be A = {a1, a2, …, a n}, construct an m×n-dimensional basic adjacency matrix . The row index of the matrix corresponds to the entity, and the column index corresponds to the attribute. In the initial state, all element values are 0, that is . To optimize the storage efficiency, adopt a sparse matrix storage structure for large-scale entity-attribute networks, and the proportion of non-zero elements 。

[0048] 2. Calculate the co-occurrence times of entities and attributes in the text data and the semantic similarity between entities and attributes respectively.

[0049] Design a sliding window mechanism to calculate the co-occurrence frequency of entities and attributes. For a text sequence T = {t1, t2, …, t L}, set the window size w (take w = 10 in the experiment), and count the number of times C i that entity e j and attribute a i,j appear simultaneously. Adopt the TF-IDF weighting strategy to enhance the key co-occurrence relationship, and define the weighted co-occurrence times as:

[0050] where |D| is the total number of documents in the corpus, and |d: e i , a j ∈ d| is the number of documents containing entity e i and attribute a j .

[0051] Adopt a hybrid embedding model to calculate the semantic similarity. First, obtain the context embedding vectors of entities and attributes through BERT-large , and then combine the structured vectors generated by domain knowledge graph embedding (such as TransE) to construct a fused representation:

[0052] where α = 0.7 is a hyperparameter. The semantic similarity is defined as: .

[0053] 3. By calculating the weighted sum of the co-occurrence times and the semantic similarity, obtain the edge weights between entities and attributes.

[0054] Propose an adaptive weighted fusion algorithm to calculate the edge weights. Define the weight function:

[0055] where λ ∈ [0, 1] is an adaptive weight parameter, which is solved through the following optimization problem:

[0056] Use the gold standard dataset for parameter calibration, and finally obtain the optimal weight λ * = 0.62.

[0057] 4. Fill the basic adjacency matrix according to the edge weights between entities and attributes.

[0058] Design a parallel filling algorithm to improve the efficiency of matrix construction. Implement distributed computing using the MapReduce framework, divide entity-attribute pairs into p data blocks, and each computing node processes m / p rows of entity data. During the filling process, introduce a threshold filtering mechanism. When W{i,j}<θ (when θ = 0.2, set M i,j = 0 to reduce noisy connections. Finally, obtain the weighted adjacency matrix M, and the distribution of its elements conforms to the power-law distribution characteristic, that is .

[0059] By constructing the adjacency matrix of entities and attributes, convert unstructured text into a computable relationship network. The adjacency matrix can be used as a knowledge storage structure that evolves over time.

[0060] In an embodiment of the present invention, based on step S3, a possible embodiment will be given below to non-restrictively elaborate on its specific implementation.

[0061] First, determine the user's concerned items. The specific methods include:[[]] (1) Obtain the keywords configured by the user and store the keywords as explicit concerned items.

[0062] Build an explicit concerned item management framework based on semantic expansion to achieve precise capture and structured storage of user needs. The system receives the original keyword set configured by the user through an interactive interface and adopts a three-layer semantic expansion strategy to improve the semantic coverage of keywords: First, perform synonym expansion based on semantic resources such as WordNet to solve the problem of diverse user expressions; second, obtain the hypernym and hyponym concepts of keywords through domain ontology construction technology to form a hierarchical semantic network; finally, use a pre-trained language model (such as BERT) to calculate the semantic similarity of word vectors and mine latent semantic related vocabulary. The expanded keyword set is stored in an inverted index structure to build a "keyword-document" mapping table, supporting fast retrieval with a time complexity of O(logn). To adapt to dynamic update requirements, design an incremental index update mechanism. When the user adds or deletes keywords, trigger a partial update of the index in an event-driven manner to ensure the real-time and efficiency of the storage structure.

[0063] (2) Obtain the user operation log, and the user operation log records the operation behaviors of the user on the historically generated tables.

[0064] Record the interaction behavior between users and tables in a five-tuple structure, including user identification, table identification, operation type, timestamp, and operation duration. The operation types cover 12 basic interaction behaviors such as viewing, filtering, sorting, and exporting. Each behavior is assigned a different semantic weight to reflect the user's attention intensity. In the log preprocessing stage, a sliding time window mechanism is adopted to focus on recent behavior patterns, and the window size is dynamically adjusted according to the user behavior stability (the default setting is 7 days). To reflect the timeliness of behaviors, an exponential decay model is introduced to decay the weights of historical operations. Operations closer to the current time are assigned higher weight coefficients. At the same time, a log cleaning mechanism is implemented to filter abnormal operation records (such as invalid operations with a duration less than 50ms), and real-time collection and transmission of logs are achieved through a distributed message queue (such as Kafka) to ensure data integrity and low-latency characteristics.

[0065] (3)Determine the implicit attention items and the attention degree of each table item according to the operation behavior; the attention degree is used to determine the front and back positions of the table items corresponding to the determined attributes in the table.

[0066] Construct a user-table-table item three-layer heterogeneous graph model. The nodes in the graph model include three types of entities: users, tables, and table items. The edge relationships include the interaction edges between users and tables (the weights are determined by the operation behavior intensity) and the inclusion edges between tables and table items (the weights are fixed values). The graph attention network (GAT) architecture is used to calculate the attention degree of table items, and the multi-head attention mechanism is used to automatically learn the association weights between different types of nodes. The attention degree calculation integrates multi-dimensional features of operation behaviors: the operation duration reflects the cognitive cost invested by the user, the operation frequency reflects the persistence of attention, and the operation recency dynamically adjusts the weights through a time decay function. To improve the generalization ability of the model, a transfer learning strategy is introduced. Based on a pre-trained model in the general domain, specific domain user behavior data is used for fine-tuning.

[0067] Then determine the attributes and table parameters to be displayed according to the user's attention items: S301. Calculate the semantic similarity between the user's attention items and attributes respectively, and mark the attributes whose semantic similarity reaches the set similarity threshold as display attributes; the semantic similarity includes explicit similarity and implicit similarity.

[0068] Explicit similarity calculation mechanism: Use a word vector model trained on a large-scale corpus to calculate the semantic distance between the keywords of the attention items and the attribute names. For example, when the user pays attention to "smartphone", the system will identify the semantic relevance between attributes such as "processor model" and "screen resolution" and this keyword. Introduce a domain knowledge graph to construct a semantic association network between entities. For example, through the knowledge graph, it can be known that "camera pixel" is a sub-attribute of "photography function". When the user pays attention to "photo-taking performance", the system can automatically associate relevant attributes.

[0069] Implicit similarity calculation mechanism: Record various operations of the user on the table (such as filtering, sorting, exporting), and analyze which attributes are frequently operated. For example, if the user filters the "price range" multiple times, it indicates that this attribute is highly relevant to the user's concerned items. Analyze the co-occurrence frequency of the user's concerned items and attributes in historical operations. For example, if the user frequently views "graphic card model" and "memory capacity" simultaneously when viewing content related to "laptop", it indicates that these two attributes have an implicit association with the user's concerned items.

[0070] Hybrid similarity fusion strategy: Calculate the confidence scores of explicit and implicit similarities respectively. For example, when keyword matching is clear and there is a strong association in the knowledge graph, the confidence score of explicit similarity is relatively high; when the user frequently operates a certain attribute recently, the confidence score of implicit similarity is relatively high. Dynamically adjust the fusion weight according to the confidence scores. For example, for new users or cold start scenarios, the weight of explicit similarity is relatively high; for old users with rich behavioral data, the weight of implicit similarity increases accordingly. Based on the distribution characteristics of the similarity scores, use the density peak clustering algorithm to automatically identify an appropriate similarity threshold. This threshold divides the attributes into three categories: highly relevant, moderately relevant, and lowly relevant.

[0071] Display attribute marking rule: For attributes whose similarity scores exceed the threshold, directly mark them as display attributes.

[0072] S302. Determine the table columns according to the display attributes, and determine the table rows according to the number of entities in the adjacency matrix.

[0073] Determine the table columns: Receive all the attribute lists after entity disambiguation processing, as well as the user's attention weights for each attribute. The user's attention weights can be obtained by the user's explicit setting (such as keyword configuration) or the system's analysis of the user's implicit behavior (such as operation logs).

[0074] Sort based on attention: First, calculate the final weight of each attribute, which is the fusion result of the user's explicit weight and the implicit weight calculated by the system. The user's explicit weight is directly taken from the user's setting. If the user does not set it, the default value 0.5 is used.

[0075] The calculation formula for the final weight is: Final weight = 0.7 * Explicit weight + 0.3 * Implicit weight. In this way, the user's explicit needs and potential interests are comprehensively considered.

[0076] To reflect the importance of the user's recent behavior, the system also introduces a time decay factor. For operations that occurred a long time ago, the corresponding attribute weights will decay over time. The decay formula is: Decayed weight = Original weight * e^(-0.1 * Number of past days).

[0077] Hierarchically process the sorted attributes and divide them into different priorities according to the attention level of the attributes. The top 3 attributes with the highest attention are selected as the main columns, and the priority is set to 1. The main columns are usually the core attributes of the entity, such as name, price, model, etc., and these attributes will be prominently displayed in the table. The next 3 attributes with medium attention are selected as secondary columns, and the priority is set to 2. The secondary columns provide supplementary information, such as rating, sales volume, size, etc., and their display positions will be secondary to the main columns. In addition to the main columns and secondary columns, the system will also add a "Details" column, which is of the type of an expandable panel. This column is used to contain all the remaining attributes to avoid layout chaos caused by too many table columns.

[0078] Adaptive column width calculation: Calculate an appropriate width for each column to ensure that the table content is clearly and beautifully displayed. The column width calculation is based on two factors: the length of the attribute name and the maximum length of the attribute value. The basic width is obtained by multiplying the length of the attribute name by 8 pixels, and the maximum content width is obtained by multiplying the maximum length in the attribute value by 6 pixels. The final column width takes the larger value of the basic width and the maximum content width, but the maximum width is limited to 200 pixels to prevent overly wide columns from affecting the overall table layout.

[0079] Determine table rows: Receive the table data after entity resolution processing and extract all unique entities from it. These entities may exist in different forms of expression, but the system needs to identify whether they belong to the same actual object.

[0080] Analyze all the extracted entities. First, based on semantic similarity, determine which entities belong to the same category. For example, for "iPhone 14" and "Apple mobile phone", calculate their semantic similarity through a pre-trained language model. When the similarity exceeds a set threshold (such as 0.8), they are determined to be entities of the same category.

[0081] In addition to semantic similarity, the system will also merge entities according to rules in a specific domain. In the field of electronic products, for different expressions such as "iPhone 14 Pro" and "Apple 14 Pro", the system will unify them into the standard expression "iPhone 14 Pro" according to the pre-set rules.

[0082] Generate a unique entity ID: The MurmurHash3 algorithm is used to generate a unique ID for the entity. This algorithm is characterized by high efficiency and low collision rate. The input for generating the ID is a string obtained by concatenating the entity type (such as "electronic product") and the core attributes (usually a combination of brand and model). This ensures that the same ID is generated for entities of the same type. When a hash conflict occurs, the system uses a chained hash table to resolve the conflict and appends an incrementing sequence number to the ID to ensure the uniqueness of each entity ID.

[0083] In one embodiment of the present invention, based on step S4, a possible embodiment will be given below to non - restrictively elaborate on its specific implementation.

[0084] Initialize the table structure according to the results of column construction and row construction: Create a table container: Generate a standard HTML table framework, including

[0085]

[0086]

[0087]

[0088] Figure 1

[0089] Figure 2

[0090]

[0091] Figure 3

[0092]

[0093]

[0094]

[0095]

[0096]

[0097]

[0098]

[0099]

[0100]

[0101]

[0102] 、 and Label; Table Header Generation: Generate table header cells according to column configurations, and set properties such as column width, alignment, etc.; Row Frame Generation: Create table rows for each entity, and pre-allocate row height and other style properties. Attribute Value Mapping and Filling: Attribute Matching Rule: Perform exact matching based on attribute names (e.g., "name" corresponds to the "Name" column); Support alias mapping configuration (e.g., "price" is mapped to the "Price" column); Implement a fuzzy matching mechanism (e.g., an attribute containing the keyword "model number" is mapped to the "Model Number" column). Data Type Processing: Numeric Type: Apply the thousand-separator and retain an appropriate number of decimal places; Date Type: Format according to user preferences (e.g., YYYY-MM-DD); Boolean Type: Convert to "Yes / No" or use an icon to represent; Array Type: Concatenate using a specific delimiter (e.g., comma, vertical bar). Value Conversion and Formatting: Currency Formatting: Add currency symbols according to the regional settings; Percentage Formatting: Automatically add a percent sign and retain two decimal places; Unit Conversion: Automatically convert weight units to common units; Text Truncation: Automatically truncate long text and add an ellipsis. In some embodiments, the artificial intelligence-based table generation system may include multiple functional modules composed of computer program segments. The computer programs of each program segment in the artificial intelligence-based table generation system may be stored in the memory of a computer device and executed by at least one processor to perform the functions of artificial intelligence-based table generation (see detailed description). In this embodiment, the artificial intelligence-based table generation system can be divided into multiple functional modules as shown. The module referred to in the present invention means a series of computer program segments that can be executed by at least one processor and can complete fixed functions, and are stored in the memory. In this embodiment, the functions of each module will be described in detail in subsequent embodiments. A first recognition module, configured to obtain text data and extract entities from the text data; A second recognition module, configured to determine the attributes of the entities through syntactic context analysis and cache the corresponding attribute values; A parameter determination module, configured to match pre-configured user concerns with the attributes, determine table columns according to the successfully matched attributes, and determine table rows according to the number of entities in the adjacency matrix; A table generation module, configured to generate a table based on the table columns and table rows, and fill the attribute values of the entities into the corresponding positions of the table. The artificial intelligence-based table generation method provided in the embodiments of the present application can be applied to devices. Those skilled in the art can understand that the device structure involved in the embodiments of the present invention does not constitute a limitation on the device. The device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements. In the embodiments of the present invention, the device includes, but is not limited to, laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers.The device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the embodiments of the present application described herein and / or claimed. Among them, the device 300 may include: a processor 310, a memory 320, and a communication unit 330. These components communicate through one or more buses. Those skilled in the art can understand that the structure of the server shown in the figure does not constitute a limitation to the present invention. It can be a bus structure, a star structure, and may also include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements. Among them, the memory 320 can be used to store the execution instructions of the processor 310. The memory 320 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk. When the execution instructions in the memory 320 are executed by the processor 310, the device 300 is enabled to execute some or all of the steps in the above method embodiments. The processor 310 is the control center of the storage device, connecting various parts of the entire electronic device through various interfaces and lines, and by running or executing the software programs and / or modules stored in the memory 320, and calling the data stored in the memory, to execute various functions of the electronic device and / or process data. The processor may be composed of an integrated circuit (IC). For example, it may be composed of a single packaged IC, or may be composed of multiple packaged ICs with the same or different functions connected together. For example, the processor 310 may only include a central processing unit (CPU). In the embodiment of the present invention, the CPU may be a single arithmetic core or may include multiple arithmetic cores. The communication unit 330 is used to establish a communication channel so that the storage device can communicate with other devices. Receive user data sent by other devices or send user data to other devices. The present invention also provides a computer storage medium. Among them, the computer storage medium may store a program, and when the program is executed, it may include some or all of the steps in the embodiments provided by the present invention. The storage medium may be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM), etc.Those skilled in the art can clearly understand that the technology in the embodiments of the present invention can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solutions in the embodiments of the present invention, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc, etc., which can store program codes, and includes several instructions for causing a computer device (which can be a personal computer, a server, or a second device, a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The same or similar parts among the various embodiments in this specification can be referred to each other. In particular, for the device embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the descriptions in the method embodiments. In several embodiments provided by the present invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are only illustrative. For example, the division of the modules is only a logical function division, and there can be other division methods in actual implementation. For example, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of systems or modules can be in electrical, mechanical, or other forms. The modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, that is, they can be located in one place, or can be distributed to multiple network modules. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. In addition, in each embodiment of the present invention, the functional modules can be integrated in a processing module, or each module can exist physically alone, or two or more modules can be integrated in one module. Although the present invention has been described in detail by referring to the drawings and in combination with the preferred embodiments, the present invention is not limited to this. Without departing from the spirit and essence of the present invention, those of ordinary skill in the art can make various equivalent modifications or substitutions to the embodiments of the present invention, and these modifications or substitutions should all be within the scope of the present invention / Any person familiar with the technical field of the present invention can easily think of changes or substitutions within the technical scope disclosed by the present invention, and they should all be covered within the protection scope of the present invention.

Claims

1. An artificial intelligence-based table generation method, characterized in that, Including: Obtain text data and extract entities from the text data; Determine the attributes of the entities through syntactic context analysis and cache the corresponding attribute values; Match the pre-configured user attention items with the attributes, determine the table columns according to the successfully matched attributes, and determine the table rows according to the number of entities in the adjacency matrix; Generate a table based on the table columns and table rows, and fill the attribute values of the entities into the corresponding positions of the table.

2. The method according to claim 1, wherein Obtain text data and extract entities from the text data, including: Obtain the text data input from the front end; Use named entity recognition technology to identify multiple entities from the text data; Perform clustering processing on the multiple entities, and merge the entities with similarity exceeding a set threshold into the same type of entities; Generate an entity list according to the merged entities.

3. The method according to claim 2, wherein Merging the entities with similarity exceeding a set threshold into the same type of entities, including: Assign a unified concept to the entity category, and replace the entity names belonging to the same entity category with the unified concept.

4. The method according to claim 1, wherein Determine the attributes of the entities through syntactic context analysis and cache the corresponding attribute values, including: Construct a syntactic dependency tree for the text data; Extract subject-verb-object triples from the syntactic dependency tree; Extract the action pairs containing entities from the subject-verb-object triples, and use the BiLSTM model to predict the attributes of the entities based on the action pairs; Locate the position of the attribute in the text data and identify the corresponding attribute value based on the position.

5. The method according to claim 1, characterized in that, The method further includes: Construct a basic adjacency matrix according to the number of entities and the number of attributes; Calculate the co-occurrence times of entities and attributes in the text data and the semantic similarity between entities and attributes respectively; Obtain the edge weights of entities and attributes by calculating the weighted sum of the co-occurrence times and the semantic similarity; Fill the basic adjacency matrix according to the edge weights of entities and attributes.

6. The method according to claim 1, characterized in that, The method further includes: Obtain the keywords configured by the user and store the keywords as explicit attention items; Obtain the user operation log, and the user operation log records the operation behaviors of the user on the historically generated table; Determine the implicit attention items and the attention degree of each table item according to the operation behaviors; The attention degree is used to determine the front and back positions of the table items corresponding to the determined attributes in the table.

7. The method according to claim 6, wherein Match the pre-configured user attention items with the attributes in the adjacency matrix, including: Calculate the semantic similarity between the user attention items and the attributes respectively, and mark the attributes with semantic similarity reaching the set similarity threshold as display attributes; the semantic similarity includes explicit similarity and implicit similarity.

8. An artificial intelligence-based table generation system, characterized in that, Including: The first recognition module is used to obtain text data and extract entities from the text data; The second recognition module is used to determine the attributes of the entities through syntactic context analysis and cache the corresponding attribute values; The parameter determination module is used to match the pre-configured user attention items with the attributes, determine the table columns according to the successfully matched attributes, and determine the table rows according to the number of entities in the adjacency matrix; The table generation module is used to generate a table based on the table columns and table rows, and fill the attribute values of the entities into the corresponding positions of the table.

9. An artificial intelligence-based form generation device, characterized in that, Including: A memory for storing a table generation program based on artificial intelligence; A processor for implementing the steps of the artificial intelligence-based table generation method according to any one of claims 1-7 when executing the artificial intelligence-based table generation program.

10. A computer-readable storage medium storing a computer program, characterized in that, An artificial intelligence-based table generation program is stored on the readable storage medium, and when the artificial intelligence-based table generation program is executed by a processor, the steps of the artificial intelligence-based table generation method according to any one of claims 1-7 are implemented.

Images