A method, device, equipment and readable storage medium for retrieving building information

Abstract

The present invention discloses a method, device, equipment and readable storage medium for searching building information, wherein the method for searching building information comprises: obtaining search information, determining the search information category corresponding to the search information; parsing the search information according to the search information category and a preset concept map, obtaining the concept entity, concept entity relationship and weight value of the concept entity corresponding to the search information; determining the search result corresponding to the search information according to the concept entity, concept entity relationship and weight value of the concept entity. By implementing the present invention, the accuracy and comprehensiveness of the search for implicit information and knowledge association are guaranteed, and the search results can be sorted according to the weight value corresponding to the concept entity, and the search results are output according to the sorting result of the weight value, which further guarantees the accuracy of the search for building information.

Description

Technical Field

[0001] The present invention relates to the technical field of information retrieval, and in particular to a method, device, equipment and readable storage medium for retrieval of building information. Background Art

[0002] The traditional search engine implementation method is to use the inverted index structure to achieve fast retrieval, implement document retrieval and sorting based on vocabulary matching based on text similarity algorithm, use a unified vocabulary and word segmentation component to process the search terms and target data, and return the matching results in order according to the similarity score. However, building materials cover a wide range, the associated knowledge between building materials is complex, and the material text description cannot fully express the implicit information of the material. For example, the implicit information of "soft sealing gate valve" is "gate valve with rubber sealing surface material", "fixed basketball stand" is "buried basketball stand", and "waterproof coating" is "coating with waterproof components, including polyurethane, asphalt, etc." At the same time, the description of building materials in the construction field lacks standard specifications. The same building material has different names in different scenes, different regions, and different data. For example, "95 bricks" is equal to "machine-made bricks", "Robinia acacia" is equal to "Robinia acacia", and "coating" includes "middle coating", "surface coating", and "paste". In addition, technical personnel in the construction field have different focuses on different building materials, and thus have different expectations and requirements for the retrieval targets of different building materials. The traditional search engine implementation methods cannot solve the problem of understanding and retrieving the implicit information expression and knowledge association of building materials in the way they process search terms and target data. Therefore, they cannot return search results as users expect. Summary of the invention

[0003] Therefore, the technical problem to be solved by the present invention is to overcome the defects of the prior art that the description of building materials is not unified and the search for building materials is not comprehensive, thereby providing a method, device, equipment and readable storage medium for retrieval of building information.

[0004] According to a first aspect, an embodiment of the present invention provides a method for retrieving building information, comprising: obtaining retrieval information, and determining a retrieval information category corresponding to the retrieval information; parsing the retrieval information according to the retrieval information category and a preset concept map to obtain concept entities, concept entity relationships, and weight values ​​of the concept entities corresponding to the retrieval information; and determining a retrieval result corresponding to the retrieval information according to the concept entities, the concept entity relationships, and the weight values ​​of the concept entities.

[0005] In combination with the first aspect, in a first implementation of the first aspect, the search information is parsed according to the search information category and a preset concept map to obtain concept entities, concept entity relationships and weight values ​​of the concept entities corresponding to the search information, including: performing entity recognition on the search information according to an entity recognition model and the search information category to determine the concept entity recognition result corresponding to the search information; proofreading the concept entity recognition result according to a concept dependency model to determine the proofread concept entities, the concept entity relationships and the weight values.

[0006] In combination with the first implementation of the first aspect, in the second implementation of the first aspect, the retrieval result corresponding to the retrieval information is determined based on the concept entity, the concept entity relationship and the weight value of the concept entity, including: constructing a query statement based on the concept entity, the concept entity relationship and the weight value of the concept entity; searching a preset index library according to the query statement to obtain the retrieval result corresponding to the retrieval information.

[0007] In combination with the first aspect, in a third implementation of the first aspect, the method for generating the concept graph includes: obtaining the retrieval information category corresponding to the retrieval information, determining the concept triple data corresponding to the retrieval information category, the concept triple data including two concept entities and the concept entity relationship between the two concept entities; calculating the weight value of each of the concept entities contained in the concept triple data; and obtaining the concept graph according to the retrieval information category, the concept triple data and the weight value.

[0008] In combination with the third implementation manner of the first aspect, in the fourth implementation manner of the first aspect, obtaining the retrieval information category corresponding to the retrieval information includes: obtaining material data in a target database; extracting concept terms from the material data in the target database to determine a concept candidate vocabulary; determining the semantic similarity of the concept candidate vocabulary based on a preset semantic model and the concept candidate vocabulary; dividing the concept candidate vocabulary based on the semantic similarity to obtain a division result; and determining the retrieval information category corresponding to the retrieval information based on the division result and the material data.

[0009] In combination with the fourth implementation manner of the first aspect, in the fifth implementation manner of the first aspect, the obtaining of the retrieval information category corresponding to the retrieval information and determining the concept triple data corresponding to the retrieval information category include: analyzing and cleaning the material data corresponding to the retrieval information category to obtain the retrieval information data; identifying the concept terms contained in the retrieval information data according to a preset extraction method to obtain the concept entities and concept entity relationships corresponding to the retrieval information data; determining the concept triple data corresponding to the retrieval information category according to the concept entities and the concept entity relationships, the concept triple data being concept entity-relationship-concept entity.

[0010] In combination with the fourth implementation of the first aspect, in the sixth implementation of the first aspect, the calculation of the weight value of each of the concept entities contained in the concept triple data includes: calculating the concept weight corresponding to the concept entity according to the frequency of occurrence of the concept entity in the material data; calculating the term weight corresponding to the concept entity according to the word frequency and inverse text frequency index of the concept term corresponding to the concept entity in the material data; determining the weight value of the concept entity according to the concept weight and the term weight.

[0011] According to the second aspect, an embodiment of the present invention provides a device for retrieving building information, including: an acquisition module, used to acquire search information and determine the search information category corresponding to the search information; a parsing module, used to parse the search information according to the search information category and a preset concept map, and obtain the concept entity, concept entity relationship and weight value of the concept entity corresponding to the search information; a determination module, used to determine the search result corresponding to the search information according to the concept entity, the concept entity relationship and the weight value of the concept entity.

[0012] According to the third aspect, an embodiment of the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method for retrieving building information described in the first aspect or any embodiment of the first aspect by executing the computer instructions.

[0013] According to a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the method for retrieving building information described in the first aspect or any implementation method of the first aspect.

[0014] The technical solution of the present invention has the following advantages:

[0015] 1. The architectural information retrieval method, device, equipment and readable storage medium provided by the present invention determine the retrieval information category according to the acquired retrieval information, parse the conceptual entities, conceptual entity relationships and weight values ​​corresponding to the conceptual entities in the retrieval information based on the constructed concept map and retrieval information category, thereby ensuring the accuracy and comprehensiveness of the retrieval of implicit information and knowledge associations, and at the same time being able to sort the search results according to the weight values ​​corresponding to the conceptual entities, and output the search results according to the sorting results of the weight values, thereby further ensuring the accuracy of the retrieval of architectural information.

[0016] 2. The building information retrieval method, device, equipment and readable storage medium provided by the present invention obtains the category and concept information corresponding to the retrieval information, generates triple data corresponding to the concept information according to the category, wherein the triple data includes two concept entities and the concept entity relationship between the two concept entities, calculates the weight value of each concept entity contained in the triple data, and generates a concept map according to the category, triple data and weight value. The unified description of the retrieval information is determined based on the acquired concept information and the concept entity relationship between the concept entities, which solves the problem of inconsistent description of building materials in the prior art. By calculating the weight value of the concept entity, different retrieval information can be ranked in importance to reflect the importance of different attributes between different retrieval information. The concept map is generated by combining the triple data generated by the weight value and the concept information, which can provide more accurate and comprehensive retrieval results for technical personnel in the construction field. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 is a flow chart of a method for retrieving building information in an embodiment of the present invention;

[0019] Figure 2 Schematic diagram of the system architecture for generating a concept map in an embodiment of the present invention;

[0020] Figure 3 is another flow chart of the method for retrieving building information in an embodiment of the present invention;

[0021] Figure 4 is another flow chart of the method for retrieving building information in an embodiment of the present invention;

[0022] Figure 5 is a flow chart of a method for generating a concept map in an embodiment of the present invention;

[0023] Figure 6 Schematic diagram of the system architecture of building material retrieval based on concept graph in an embodiment of the present invention;

[0024] Figure 7 is another flow chart of a method for generating a concept map in an embodiment of the present invention;

[0025] Figure 8 is another flow chart of a method for generating a concept map in an embodiment of the present invention;

[0026] Fig. 9 is a schematic diagram of a concept map in an embodiment of the present invention;

[0027] Fig.10 It is a principle block diagram of a device for retrieving building information in an embodiment of the present invention;

[0028] Fig.11 Schematic diagram of the structure of a computer device in an embodiment of the present invention. DETAILED DESCRIPTION

[0029] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0030] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.

[0031] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, it can also be the internal connection of two components, it can be a wireless connection, or it can be a wired connection. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0032] In addition, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0033] Example 1

[0034] This embodiment provides a method for retrieving building information, which searches for building materials based on concept graphs, avoids the problem of inaccurate and incomplete building material searches caused by inconsistent building material descriptions, and can return search results in a reasonable order expected by users while ensuring the accuracy and comprehensiveness of search results. Figure 1 As shown, the building information retrieval method includes the following steps:

[0035] S11, obtaining search information, and determining the search information category corresponding to the search information.

[0036] Exemplarily, the search information is a concept entity corresponding to the search information input by the user, and the concept entity may be composed of multiple concept terms, and the concept terms are nouns that constitute the search information. According to the concept entity contained in the search information, the search information category corresponding to the search information can be determined. Since the concept terms corresponding to different search information may correspond to different concept types in different material data descriptions, that is, a polysemy problem, for example: "stainless steel" is a "material" concept in "stainless steel plate" and a "subject" concept in "304 stainless steel". The search information category corresponding to the search information is determined by a machine learning algorithm. The specific implementation method is: manually classify and annotate a large amount of material data description text information containing concept terms to obtain a classifier training corpus, and use a convolutional neural network (CNN) algorithm to train the model to obtain a classification model based on the CNN algorithm. In the process of searching, the CNN classification model can be used to predict the search information category of the search information. For example, if the search information is "stainless steel plate", the search information can be "stainless steel" and "plate". According to the search information "stainless steel" and "plate", the search information category "ferrous and non-ferrous metals" corresponding to the search information can be determined. Among them, the training methods of classification models include but are not limited to machine learning models, deep learning models, syntactic dependency methods, and vocabulary matching methods.

[0037] S12, parsing the search information according to the search information category and the preset concept map to obtain the concept entities, concept entity relationships and concept entity weight values ​​corresponding to the search information.

[0038] For example, the preset concept graph can be constructed and generated based on the categories of building materials from the database, the concept entities corresponding to the building materials, the concept entity relationships corresponding to the building materials, and the weight values ​​between the concept entities corresponding to the building materials. Based on the constructed concept graph, the concept entities, concept entity relationships, and weight values ​​of the concept entities corresponding to the search information category can be matched in the preset concept graph according to the search information category. Figure 2 As shown, the system architecture diagram of building material search based on concept graph may include: text classification, entity recognition, graph linking, query statement construction, data retrieval and other modules.

[0039] S13, determining the search result corresponding to the search information according to the concept entity, the concept entity relationship and the weight value of the concept entity.

[0040] Exemplarily, according to the concept entities, concept entity relationships and weight values ​​of the concept entities of the corresponding search information matched in the preset concept graph, multiple matching results can be obtained. The matching results are sorted from high to low according to the weight values ​​of the concept entities to obtain the search results corresponding to the search information. For example, according to the concept entities, concept entity relationships and weight values ​​of the concept entities of the corresponding search information matched in the preset concept graph, matching material data 1, matching material data 2 and matching material data 3 are obtained, and the weight values ​​corresponding to the matching material data are 0.9, 0.2 and 0.6, then the search results corresponding to the search information are: matching material data 1, matching material data 3 and matching material data 2.

[0041] The architectural information retrieval method provided in this embodiment determines the retrieval information category based on the acquired retrieval information, and parses the concept entities, concept entity relationships and weight values ​​corresponding to the concept entities in the retrieval information based on the constructed concept graph and the retrieval information category, thereby ensuring the accuracy and comprehensiveness of the retrieval of implicit information and knowledge associations, and at the same time being able to sort the search results according to the weight values ​​corresponding to the concept entities, and output the search results according to the sorting results of the weight values, thereby improving the accuracy of the retrieval of architectural information and ensuring that the search results are returned in a reasonable order expected by the user.

[0042] As an optional implementation, Figure 3 As shown, the above step S12 includes:

[0043] S121, performing entity recognition on the search information according to the search information category and the entity recognition model, and obtaining a concept entity recognition result corresponding to the search information.

[0044] Exemplarily, in order to improve the accuracy of concept entity recognition, a corresponding concept entity recognition training set is annotated for each retrieval information, and a machine learning model of long short-term memory network (Long Short-Term Memory, LSTM) + conditional random field (conditional random field, CRF) is used for training to obtain an entity recognition model. According to the determined retrieval information category, the corresponding LSTM+CRF model is selected to perform entity recognition on the retrieval information to obtain a concept entity recognition result set of the concept terms corresponding to the retrieval information. The training method of the entity recognition model includes but is not limited to a machine learning model, a deep learning model, a syntactic dependency method, and a vocabulary matching method.

[0045] S122, proofreading the concept entity recognition result based on the concept dependency model, and determining the proofread concept entity, concept entity relationship and weight value.

[0046] Exemplarily, the concept entity recognition result set obtained in step S121 is proofread and corrected through a concept dependency model, wherein the concept dependency is a common sense constraint rule for the combination of different concept information of the building retrieval information, for example: "stainless steel door fire extinguisher cabinet", the entity recognition model may recognize "stainless steel door, entity" or "fire extinguisher cabinet, entity", however, it is impossible for two ontologies to exist in the same retrieval information, then "stainless steel door, entity" can be changed to "stainless steel door, composition", and then the concept entity data set related to the concept entity can be queried in the NEO4J graph database to determine the concept entity corresponding to the retrieval information, the concept entity relationship and the weight value of the concept entity.

[0047] As an optional implementation, Figure 4 As shown, the above step S13 includes:

[0048] S131, constructing a query statement according to the concept entities, the concept entity relationships and the weight values ​​of the concept entities.

[0049] Exemplarily, the search engine query statement is constructed in combination according to the concept information and weight values ​​of the concept entities in the obtained concept entity data set. For example, for the solr search engine, if the search information is "304 stainless steel plate", the concept entity and weight value corresponding to the search information are "name: 304, type: parameter, weight: 10", "name: stainless steel, type: material, weight: 20", "plate, object", and the solr search engine query statement can be constructed as: "(material: stainless steel) AND (object: plate)".

[0050] S132, searching a preset index library according to the query statement to obtain search results corresponding to the search information.

[0051] Exemplarily, the preset index library stores conceptual entities corresponding to the search information, and the search information is searched in the preset index library according to the query statement to obtain the search results corresponding to the search information. For example, the search information is searched on the Solr material data index library through the Solr search engine query statement to obtain the search results corresponding to the search information, and the search results can be returned to the user for viewing.

[0052] As an optional implementation, Figure 5 As shown, the method for generating a concept map may include the following steps:

[0053] S21, obtaining a search information category corresponding to the search information, and determining concept triple data corresponding to the search information category, wherein the concept triple data includes two concept entities and a concept entity relationship between the two concept entities.

[0054] Exemplarily, the retrieval information is building materials in the construction field, such as stainless steel plates; the retrieval information category is the type of building materials, such as ferrous and non-ferrous metals. Concept entities can be defined according to building material experts, including but not limited to "unit", "shape", "characteristic", "ingredient", "function", "color", "installation", "location", "origin", "attachment", "composition", "target", "parameter", "artifact", and the concept entity relationship can be represented as a directed edge connecting the concept entities in the concept graph. The concept entity relationship type can include but is not limited to "include", "belong to", "equal to", and "mutually exclusive". According to the category of the retrieval information, triple data is extracted from the concept entities and concept entity relationships corresponding to the retrieval information, and the triple data includes two concept entities and the concept entity relationship between the two concept entities.

[0055] S22, calculating the weight value of each concept entity included in the concept triple data.

[0056] Exemplarily, based on the extracted triple data, the weight value corresponding to the concept entity contained in the triple data is calculated. The weight value can be composed of two parts: the concept weight and the term weight corresponding to the concept entity. The concept weight value and the term weight value are calculated separately.

[0057] S23, obtaining a concept map according to the search information category, concept triple data and weight value.

[0058] Exemplarily, the material data corresponding to the search information is analyzed and cleaned, the search information category, concept entity and concept entity relationship corresponding to the search information are obtained, the concept triple data corresponding to the search information is extracted from the concept entity and the concept entity relationship, and the weight value corresponding to the concept entity is calculated, such as Figure 6 As shown, the system architecture for generating the concept map may include: concept term extraction, concept word library classification, concept definition, map construction and other modules, and the search information categories, concept entity relationships and weight values ​​corresponding to the search information are stored in the neo4j graph database to generate the concept map, so as to avoid the problem of inaccurate and incomplete search of building materials due to inconsistent descriptions of building materials. The storage of the search information categories, concept entity relationships and weight values ​​corresponding to the search information includes but is not limited to graph databases, relational databases and nosql databases.

[0059] By obtaining the search information category corresponding to the search information, generating the concept triple data corresponding to the search information according to the search information category, calculating the weight value of each concept entity contained in the concept triple data, and generating a concept map according to the search information category, the concept triple data and the weight value. The concept map determines the unified description of the search information according to the acquired concept entities and the concept entity relationship between the concept entities, which solves the problem of inconsistent description of building materials in the prior art. By calculating the weight value of the concept entity, the importance of different search information can be ranked to reflect the importance of different attributes between different search information. The concept triple data determined by combining the weight value, the concept entity and the concept entity relationship can generate a concept map, which can provide more accurate and comprehensive search results for technical personnel in the construction field.

[0060] As an optional implementation, Figure 7 As shown, in the above step S21, the process of obtaining the category and concept information corresponding to the search information specifically includes:

[0061] S211, obtaining material data in the target database.

[0062] Exemplarily, the target database includes material data from different sources, such as behavioral data and encyclopedia data, which is not limited in the present application. The data in the target database is analyzed to obtain material data corresponding to the search information.

[0063] S212, extracting concept terms from the material data in the target database to generate a concept candidate word library.

[0064] Exemplarily, an unsupervised word segmentation algorithm is used to extract concept terms from the material description data of existing building materials in the target database. For example, in order to complete the expression of the concept terms of the search information, "hot-dip galvanized steel pipe" can be split into "hot-dip galvanized", "steel", and "pipe". Considering that the concept terms have the characteristic of repeating in a relatively stable period, this embodiment uses an unsupervised word segmentation algorithm to extract concept terms from the description data of the existing search information, and statistically analyzes the lexical co-occurrence probability and random combination probability of the concept terms. Combined with the lexical occurrence frequency of the concept terms, the basic vocabulary corresponding to the concept terms is extracted, and then the Long March words in the basic vocabulary are split using the short words in the basic vocabulary. For example, if the basic vocabulary contains "galvanized steel pipe", "galvanized", "steel pipe", "steel", and "pipe", the splittable "galvanized steel pipe" and "steel pipe" are removed to ensure that the information expression of the concept terms is a basic information unit granularity with a singleness, and the acquired concept terms are combined to generate a basic concept candidate vocabulary.

[0065] S213, determining the semantic similarity of the concept candidate word library according to the preset semantic model and the concept candidate word library.

[0066] Exemplarily, the preset semantic model may be a word vector semantic model word2vec, through which the semantic similarity of the concept terms included in the concept candidate word library is obtained. This application does not limit the preset semantic model, and those skilled in the art may determine it according to actual needs.

[0067] S214, dividing the concept candidate word library according to semantic similarity to obtain a division result.

[0068] Exemplarily, by using the clustering algorithm k-means and semantic similarity to perform conceptual division on the concept terms in the concept candidate word library, the accuracy of the obtained concept candidate word library division results can be ensured. For example, the concept term "stainless steel plate" in the concept candidate word library can obtain the division results "stainless steel" and "plate". This application does not limit the clustering algorithm used for division, and those skilled in the art can determine it according to actual needs.

[0069] S215, determining the search information category corresponding to the search information according to the classification result and the material data.

[0070] For example, in order to reduce the error between the concept division produced by the manual summary definition method and the natural law of the retrieval information composition, the machine learning method can be used to automatically perform semantic division on the concept candidate word library by calculating semantic similarity, and then analyze the conceptual commonality of the vocabulary expression information of the concept terms in different clusters, and give the concept definition and explanation of each concept term cluster. For example: Concept term cluster 1 contains the words "stainless steel", "polyurethane", "PVC", and "aluminum alloy", which are all terms expressing materials. The concept definition of this cluster is given as "concept definition: material; concept explanation: the main chemical composition of the material". The division results obtained by the clustering algorithm and the material data contained in the target database can be used to determine the category and concept information corresponding to the retrieval information. For example, the category corresponding to the retrieval information "stainless steel plate" is "ferrous and non-ferrous metals", the concept entities corresponding to the retrieval information "stainless steel plate" are "stainless steel" and "plate", and the concept entity relationship is "concept feature".

[0071] The composition characteristics of architectural retrieval information are the main information contained in the architectural retrieval information description text, such as shape, form, function, material, etc. Its characteristics can be summarized by analyzing a large amount of material data, or can be the professional experience of experts in related fields. Concept information is defined in combination with the composition characteristics of architectural retrieval information to ensure the rationality, universality and interpretability of the concept information definition in the technical field corresponding to the retrieval information.

[0072] As an optional implementation, Figure 7 As shown, the process of determining the concept triple data corresponding to the search information category in the above step S21 specifically includes:

[0073] S216, analyzing and cleaning the material data corresponding to the search information category to obtain search information data.

[0074] Exemplarily, under the category corresponding to the search information, material data corresponding to the search information is obtained, and the material data corresponding to the category is analyzed and cleaned to remove meaningless conceptual terms, thereby obtaining accurate and comprehensive search information data.

[0075] S217, based on a preset extraction method, identifying the concept terms contained in the search information data, and obtaining the concept entities and concept entity relationships corresponding to the search information data.

[0076] Exemplarily, the concept terms contained in the search information data are identified, the concept terms corresponding to the search information are extracted from the search information data according to a preset extraction method, the concept entities and concept entity relationships of the acquired concept terms are distinguished, and then the concept entities and concept entity relationships corresponding to the search information data are obtained. Optionally, the preset extraction method includes but is not limited to lexical patterns, co-occurrence analysis, text clustering and deep learning.

[0077] S218, determining the concept triple data corresponding to the search information category according to the concept entity and the concept entity relationship, the concept triple data is concept entity-relationship-concept entity.

[0078] Exemplarily, the identified conceptual entities and conceptual entity relationships are combined into conceptual triple data, wherein the conceptual triple data includes two groups of conceptual entities and one group of conceptual entity relationships, namely, conceptual entity-relationship-conceptual entity. The method for extracting conceptual triple data is mainly to identify the conceptual terms corresponding to the conceptual entities contained in the retrieval information description under the current category. For example, category: ferrous and non-ferrous metals, material: stainless steel plate, the recognition results of the conceptual entity are: "stainless steel, material", "plate, object", and then the triple data is extracted through the concept dependency relationship model verification, and the triple data is (ferrous and non-ferrous metals-concept features-stainless steel), (ferrous and non-ferrous metals-concept features-plate). Among them, the conceptual entity can include attributes such as conceptual entity type and conceptual entity weight, and the conceptual entity relationship can be determined using word vector similarity, co-occurrence analysis, expert system and other methods.

[0079] As an optional implementation, Figure 8 As shown, the above step S22 includes:

[0080] S221, calculating the concept weight corresponding to the concept entity according to the frequency of occurrence of the concept entity in the material data.

[0081] Exemplarily, in order to avoid the problem of uneven distribution of concept terms, the weight value can be composed of two parts: concept weight and vocabulary weight. Among them, the concept weight can be obtained by multiplying 10 according to the sorting order of the frequency of concept entities in the accumulated user search term data from high to low. For example, the material data containing concept terms corresponding to concept entities in the target database are material data 1, material data 2 and material data 3, and the frequency of occurrence of concept terms corresponding to concept entities in material data is 0.2, 0.1 and 0.3 respectively, then the concept weights are 2, 1 and 3.

[0082] S222, calculating the term weight corresponding to the concept entity according to the term frequency and inverse text frequency index of the concept term corresponding to the concept entity in the material data.

[0083] Exemplarily, the term weight can be determined based on the weighted value TF-IDF of the concept term corresponding to the concept entity in the entire target database, and the weighted value TF-IDF can be obtained by multiplying the term frequency value by the inverse document frequency value. The term frequency value TF can be determined based on the frequency of the concept term corresponding to the concept entity in the material data. For example, the number of occurrences of the concept term "stainless steel" in the material data is 200, and the total number of terms contained in the material data is 20,000, then the term frequency value TF = 2000 / 20000 = 0.1. The inverse document frequency value IDF can be obtained by dividing the total number of material data by the number of material data containing the concept term, and then taking the logarithm of the quotient with a base of 10. For example, the number of material data containing the concept term "stainless steel" in the target database is 100, and the total number of material data contained in the target database is 10,000, where the number of material data can be the number of files, and the inverse document frequency value IDF = lg (10000 / 100) = 2, then the term weight value = 0.1*2 = 0.2.

[0084] S223, determining the weight value of the concept entity according to the concept weight and the term weight.

[0085] Exemplarily, the obtained concept weight is multiplied by the term weight to obtain the weight value corresponding to the concept entity. For example, the frequency of occurrence of the concept term corresponding to the concept entity in material data 1 is 0.2, the term frequency value TF of the concept term corresponding to the concept entity in the material data is 0.1, and the inverse document frequency value IDF is 2, then the weight value of the concept entity in material data 1 is 0.4.

[0086] As Fig. 9 As shown in the figure, for example, a concept map can be constructed according to the generation method of the concept map. The building concept map can contain more than 20 types of building material concept information, more than 100,000 concept entities, and cover 30 primary categories of building materials. The building material concept map brings together various concept entities involved in different categories of building materials and the relationships between various concept entities. Through the deconstruction of concept entities and relationships, a more comprehensive understanding of the information contained in building materials, their implicit information and extended relationships can be achieved. Applying the generation method of the concept map to the building material search scenario helps to improve the precision and recall rates.

[0087] Example 2

[0088] This embodiment provides a building information retrieval device, which searches for building materials based on the concept map generation device in the above embodiment, avoiding the problem of inaccurate and incomplete building material searches caused by inconsistent building material descriptions. On the basis of ensuring the accuracy and comprehensiveness of the search results, the search results can be returned in a reasonable order expected by the user, such as Fig.10 As shown, the device comprises:

[0089] The acquisition module 21 is used to acquire the search information and determine the search information category corresponding to the search information. For details, please refer to the relevant description of the corresponding step S11 in the above method embodiment, which will not be repeated here.

[0090] The parsing module 22 is used to parse the search information according to the search information category and the preset concept map to obtain the concept entity, concept entity relationship and concept entity weight value corresponding to the search information. For details, please refer to the relevant description of step S12 in the above method embodiment, which will not be repeated here.

[0091] The determination module 23 is used to determine the search result corresponding to the search information according to the concept entity, the concept entity relationship and the weight value of the concept entity. For details, please refer to the relevant description of the corresponding step S13 of the above method embodiment, which will not be repeated here.

[0092] The building information retrieval device provided in this embodiment determines the retrieval information category based on the acquired retrieval information, and parses the concept entities, concept entity relationships and weight values ​​corresponding to the concept entities in the retrieval information based on the constructed concept graph and the retrieval information category, thereby ensuring the accuracy and comprehensiveness of the retrieval of implicit information and knowledge associations, and at the same time can sort the search results according to the weight values ​​corresponding to the concept entities, and output the search results according to the sorting results of the weight values, thereby further ensuring the retrieval accuracy of the building information.

[0093] As an optional implementation, the above-mentioned parsing module 22 includes:

[0094] The first recognition submodule is used to perform entity recognition on the search information according to the entity recognition model and the search information category, and determine the concept entity recognition result corresponding to the search information. For details, please refer to the relevant description of step S121 in the above method embodiment, which will not be repeated here.

[0095] The proofreading submodule is used to proofread the concept entity recognition results according to the concept dependency model, and determine the proofread concept entities, concept entity relationships and weight values. For details, please refer to the relevant description of step S122 of the above method embodiment, which will not be repeated here.

[0096] As an optional implementation, the determination module 23 includes:

[0097] The construction submodule is used to construct a query statement according to the concept entity, the concept entity relationship and the weight value of the concept entity. For details, please refer to the relevant description of the corresponding step S131 of the above method embodiment, which will not be repeated here.

[0098] The search submodule is used to search the preset index library according to the query statement to obtain the search results corresponding to the search information. For details, please refer to the relevant description of the corresponding step S132 of the above method embodiment, which will not be repeated here.

[0099] As an optional implementation, the above-mentioned parsing module 22 also includes a concept map generation submodule, and the concept map generation submodule includes:

[0100] The acquisition unit is used to acquire the search information category corresponding to the search information, determine the concept triple data corresponding to the search information category, and the concept triple data includes two concept entities and the concept entity relationship between the two concept entities. For details, please refer to the relevant description of the corresponding step S21 of the above method embodiment, which will not be repeated here.

[0101] The calculation unit is used to calculate the weight value of each concept entity contained in the concept triple data. For details, please refer to the relevant description of the corresponding step S22 of the above method embodiment, which will not be repeated here.

[0102] The generating unit is used to obtain a concept map according to the search information category, the concept triple data and the weight value. For details, please refer to the relevant description of the corresponding step S23 of the above method embodiment, which will not be repeated here.

[0103] By obtaining the search information category corresponding to the search information, generating the concept triple data corresponding to the search information according to the search information category, calculating the weight value of each concept entity contained in the concept triple data, and generating a concept map according to the search information category, the concept triple data and the weight value. The concept map determines the unified description of the search information according to the acquired concept entities and the concept entity relationship between the concept entities, which solves the problem of inconsistent description of building materials in the prior art. By calculating the weight value of the concept entity, the importance of different search information can be ranked to reflect the importance of different attributes between different search information. The concept triple data determined by combining the weight value, the concept entity and the concept entity relationship can generate a concept map, which can provide more accurate and comprehensive search results for technical personnel in the construction field.

[0104] As an optional implementation, the acquisition unit includes:

[0105] The data acquisition subunit is used to acquire the material data in the target database. For details, please refer to the relevant description of the corresponding step S211 in the above method embodiment, which will not be repeated here.

[0106] The extraction subunit is used to extract concept terms from the material data in the target database and determine the concept candidate word library. For details, please refer to the relevant description of the corresponding step S212 in the above method embodiment, which will not be repeated here.

[0107] The first determination subunit is used to determine the semantic similarity of the concept candidate word library according to the preset semantic model and the concept candidate word library. For details, please refer to the relevant description of the corresponding step S213 of the above method embodiment, which will not be repeated here.

[0108] The division subunit is used to divide the concept candidate word library according to the semantic similarity to obtain the division result. For details, please refer to the relevant description of the corresponding step S214 of the above method embodiment, which will not be repeated here.

[0109] The second determination subunit is used to determine the search information category corresponding to the search information according to the classification result and the material data. For details, please refer to the relevant description of the corresponding step S215 of the above method embodiment, which will not be repeated here.

[0110] As an optional implementation, the acquisition unit further includes:

[0111] The analysis subunit is used to analyze and clean the material data corresponding to the search information category to obtain the search information data. For details, please refer to the relevant description of the corresponding step S216 of the above method embodiment, which will not be repeated here.

[0112] The identification subunit is used to identify the concept terms contained in the search information data according to the preset extraction method, and obtain the concept entities and concept entity relationships corresponding to the search information data. For details, please refer to the relevant description of step S217 of the above method embodiment, which will not be repeated here.

[0113] The generating subunit is used to determine the concept triple data corresponding to the search information category according to the concept entity and the concept entity relationship, and the concept triple data is concept entity-relationship-concept entity. For details, please refer to the relevant description of step S218 corresponding to the above method embodiment, which will not be repeated here.

[0114] As an optional implementation, the above-mentioned calculation unit includes:

[0115] The first calculation subunit is used to calculate the concept weight corresponding to the concept entity according to the occurrence frequency of the concept entity in the material data. For details, please refer to the relevant description of the corresponding step S221 of the above method embodiment, which will not be repeated here.

[0116] The second calculation subunit is used to calculate the weight of the term corresponding to the concept entity according to the word frequency and inverse text frequency index of the concept term corresponding to the concept entity in the material data. For details, please refer to the relevant description of step S222 of the above method embodiment, which will not be repeated here.

[0117] The third determination subunit is used to determine the weight value of the concept entity according to the concept weight and the term weight. For details, please refer to the relevant description of the corresponding step S223 of the above method embodiment, which will not be repeated here.

[0118] Example 3

[0119] The embodiment of the present invention also provides a computer device, such as Fig.11 As shown, the device includes a processor 31 and a memory 32, wherein the processor 31 and the memory 32 can be connected via a bus or other means. Fig.11 The example of connecting through bus is taken in the following.

[0120] The processor 31 may be a central processing unit (CPU). The processor 31 may also be other general-purpose processors, digital signal processors (DSP), graphics processors (GPU), embedded neural network processors (NPU), or other dedicated deep learning coprocessors, application specific integrated circuits (ASIC), field programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or a combination of the above chips.

[0121] The memory 32 is a non-transitory computer-readable storage medium that can be used to store non-transitory software programs, non-transitory computer executable programs and modules, such as program instructions / modules corresponding to the building information retrieval method in the embodiment of the present invention (e.g. Fig.10 The processor 31 executes various functional applications and data processing of the processor by running the non-transient software programs, instructions and modules stored in the memory 32, that is, the building information retrieval method in the above method embodiment is implemented.

[0122] The memory 32 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and applications required for at least one function; the data storage area may store data created by the processor 31, etc. In addition, the memory 32 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage devices. In some embodiments, the memory 32 may optionally include a memory remotely arranged relative to the processor 31, and these remote memories may be connected to the processor 31 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0123] The one or more modules are stored in the memory 32 and when executed by the processor 31, perform the following steps: Figure 1-Figure 9 The method for retrieving building information in the illustrated embodiment.

[0124] By obtaining the category and concept information corresponding to the search information, triple data corresponding to the concept information is generated according to the category, wherein the triple data includes two concept entities and the concept entity relationship between the two concept entities, the weight value of each concept entity contained in the triple data is calculated, and a concept map is generated according to the category, triple data and weight value. The unified description of the search information is determined according to the acquired concept information and the concept entity relationship between the concept entities, which solves the problem of inconsistent description of building materials in the prior art. By calculating the weight value of the concept entity, the importance of different search information can be ranked to reflect the importance of different attributes between different search information. The concept map is generated by combining the triple data generated by the weight value and the concept information, which can provide more accurate and comprehensive search results for technical personnel in the construction field.

[0125] The retrieval information category is determined based on the acquired retrieval information, and the concept entities, concept entity relationships and weight values ​​corresponding to the concept entities in the retrieval information are parsed based on the constructed concept map and retrieval information category, thereby ensuring the accuracy and comprehensiveness of the retrieval of implicit information and knowledge associations. At the same time, the search results can be sorted according to the weight values ​​corresponding to the concept entities, and the search results are output according to the sorting results of the weight values, further ensuring the accuracy of building information retrieval.

[0126] For details of the above computer equipment, please refer to Figures 1 to 10 The corresponding related descriptions and effects in the illustrated embodiments can be understood and will not be repeated here.

[0127] The embodiment of the present invention further provides a non-transitory computer storage medium, wherein the computer storage medium stores computer executable instructions, and the computer executable instructions can execute the building information retrieval method in any of the above method embodiments. The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), a random access memory (RAM), a flash memory, a hard disk drive (HDD) or a solid-state drive (SSD), etc.; the storage medium can also include a combination of the above types of memory.

[0128] Obviously, the above embodiments are merely examples for the purpose of clear explanation, and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived therefrom are still within the scope of protection of the invention.

Claims

1. A method for retrieving building information, characterized in that: include: Acquire search information, and determine the search information category corresponding to the search information; According to the search information category and the preset concept map, the search information is parsed to obtain the concept entity, concept entity relationship and weight value of the concept entity corresponding to the search information, wherein the method for generating the concept map includes: Acquire the search information category corresponding to the search information, determine the concept triple data corresponding to the search information category, the concept triple data including two concept entities and a concept entity relationship between the two concept entities; Calculating the weight value of each of the concept entities contained in the concept triple data; Obtaining the concept map according to the search information category, the concept triple data and the weight value; Determine the search result corresponding to the search information according to the concept entity, the concept entity relationship and the weight value of the concept entity Wherein, the obtaining of the search information category corresponding to the search information includes: Acquire material data in the target database; Extracting concept terms from the material data in the target database to determine a concept candidate word library; Determining the semantic similarity of the concept candidate word library according to a preset semantic model and the concept candidate word library; Dividing the concept candidate word library according to the semantic similarity to obtain a division result; Determining the search information category corresponding to the search information according to the division result and the material data; The obtaining of the search information category corresponding to the search information and determining the concept triple data corresponding to the search information category includes: Analyzing and cleaning the material data corresponding to the search information category to obtain search information data; According to a preset extraction method, the concept terms contained in the search information data are identified to obtain the concept entities and concept entity relationships corresponding to the search information data; According to the concept entity and the concept entity relationship, the concept triple data corresponding to the search information category is determined, and the concept triple data is concept entity-relationship-concept entity.

2. The method according to claim 1, characterized in that The step of parsing the search information according to the search information category and the preset concept map to obtain the concept entity, concept entity relationship and weight value of the concept entity corresponding to the search information includes: Perform entity recognition on the search information according to the entity recognition model and the search information category, and determine the concept entity recognition result corresponding to the search information; The concept entity recognition result is proofread according to the concept dependency model to determine the proofread concept entity, the concept entity relationship and the weight value.

3. The method according to claim 2, characterized in that Determining the search result corresponding to the search information according to the concept entity, the concept entity relationship and the weight value of the concept entity includes: Constructing a query statement according to the concept entities, the concept entity relationships and the weight values ​​of the concept entities; The preset index library is searched according to the query statement to obtain the search results corresponding to the search information.

4. The method according to claim 1, characterized in that: The calculating the weight value of each of the concept entities contained in the concept triple data includes: Calculate the concept weight corresponding to the concept entity according to the occurrence frequency of the concept entity in the material data; Calculate the term weight corresponding to the concept entity according to the term frequency and inverse text frequency index of the concept term corresponding to the concept entity in the material data; The weight value of the concept entity is determined according to the concept weight and the term weight.

5. A building information retrieval device, characterized in that: include: An acquisition module, used to acquire search information and determine the search information category corresponding to the search information; A parsing module is used to parse the search information according to the search information category and a preset concept map, and obtain the concept entity, concept entity relationship and weight value of the concept entity corresponding to the search information, wherein the method for generating the concept map includes: obtaining the search information category corresponding to the search information, determining the concept triple data corresponding to the search information category, wherein the concept triple data includes two concept entities and the concept entity relationship between the two concept entities; calculating the weight value of each concept entity contained in the concept triple data; obtaining the concept map according to the search information category, the concept triple data and the weight value, wherein obtaining the search information category corresponding to the search information includes: obtaining material data in a target database; extracting concept terms from the material data in the target database, and determining a concept candidate word library; According to a preset semantic model and the concept candidate word library, the semantic similarity of the concept candidate word library is determined; according to the semantic similarity, the concept candidate word library is divided to obtain a division result; according to the division result and the material data, the retrieval information category corresponding to the retrieval information is determined; the retrieval information category corresponding to the retrieval information is obtained, and the concept triple data corresponding to the retrieval information category is determined, including: analyzing and cleaning the material data corresponding to the retrieval information category to obtain retrieval information data; according to a preset extraction method, the concept terms contained in the retrieval information data are identified to obtain the concept entities and concept entity relationships corresponding to the retrieval information data; according to the concept entities and the concept entity relationships, the concept triple data corresponding to the retrieval information category is determined, and the concept triple data is concept entity-relationship-concept entity; The determination module is used to determine the search result corresponding to the search information according to the concept entity, the concept entity relationship and the weight value of the concept entity.

6. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the building information retrieval method according to any one of claims 1 to 4 by executing the computer instructions.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the building information retrieval method according to any one of claims 1 to 4.

Images