Search method, device and electronic device based on knowledge graph

Abstract

The present application discloses a search method, device, electronic device and computer-readable storage medium based on a knowledge graph, the method comprising: obtaining a first query text; determining at least one first information entity associated with the first query text, screening out a second information entity associated with the first information entity from a pre-constructed knowledge graph, determining from the knowledge graph the information corresponding to each information entity and the association between each information entity, and determining the display method of each information entity, the information corresponding to each information entity and the association between each information entity according to a predetermined rule, and providing them to a display interface that needs to be displayed. The solution provided by the present application enables planners to no longer need to repeatedly search multiple times, nor to comb through the searched content too much, thereby simplifying the information search process before planners edit information, thereby improving the efficiency of planners in editing information.

Description

Technical Field

[0001] The present application relates to the field of information search technology, and specifically to a search method, device, electronic device and computer-readable storage medium based on a knowledge graph. Background Art

[0002] Information is an important way for people to obtain information. Planners need to plan and edit it, and publish the edited information through the Internet, television, newspapers, etc., so that people can consult it. When planning information manuscripts, planners often need to search for the required information from the information base first, for example, search for relevant information, related papers, etc., and then sort out the searched information to edit the required information manuscripts based on the sorted information.

[0003] With the rapid development of information dissemination technology and the popularization of Internet technology, the growth rate of information is very rapid, so the amount of information in the information base and the real-time increment are very large. The amount of relevant information searched by planners from a large amount of information base is usually very large, and the process of sorting is also very cumbersome. Therefore, the search and sorting process will consume a lot of time and energy of planners, resulting in low efficiency of information editing by planners. Summary of the invention

[0004] This application provides a search method, device, electronic device and computer-readable storage medium based on knowledge graph, which can quickly search and sort out the information required by planners, thereby improving the efficiency of planners in editing information. The specific scheme is as follows:

[0005] In a first aspect, the present application provides a search method based on a knowledge graph, which is applied to a server, and the method comprises:

[0006] Get the first query text;

[0007] Determine at least one first information entity associated with the first query text, where the first information entity is used to represent an information entity involved in the information corresponding to the first query text;

[0008] Filtering at least one second information entity associated with the first information entity from a pre-constructed knowledge graph;

[0009] Determine from the knowledge graph the information corresponding to each information entity and the association relationship between each information entity, wherein each information entity includes each first information entity and each second information entity;

[0010] The display mode of each information entity, the information corresponding to each information entity, and the association relationship between each information entity is determined according to a predetermined rule, and provided to a display interface that needs to be displayed.

[0011] Optionally, determining a first information entity associated with the first query text includes:

[0012] Performing entity recognition on the query text first query text to obtain at least one third information entity;

[0013] A first information entity associated with the first query text is determined according to each of the third information entities.

[0014] Optionally, determining the first information entity associated with the first query text according to each of the third information entities includes:

[0015] Performing entity expansion on the third information entity based on a pre-built knowledge graph to obtain at least one expanded entity associated with the third information entity;

[0016] A first information entity associated with the query text first query text is determined according to the third information entity and the extended entity.

[0017] Optionally, before obtaining the first query text, the method further includes:

[0018] Get the second query text;

[0019] Searching for target information matching the second query text from each piece of information stored in the pre-built knowledge graph;

[0020] Determine each information entity associated with the target information from the knowledge graph;

[0021] The entity expansion of the third information entity based on the pre-built knowledge graph includes:

[0022] Performing entity expansion on the third information entity based on each information entity associated with the target information;

[0023] The first query text is a text obtained by expanding the second query text.

[0024] Optionally, the knowledge graph stores stored information and events, locations and people corresponding to the stored information;

[0025] The determining, from the knowledge graph, each information entity associated with the target information includes:

[0026] Determining each target event involved in the target information;

[0027] Searching for each target occurrence location and each target person associated with the target event from the knowledge graph;

[0028] The target event, the target occurrence location, and the target person are determined as information entities associated with the target information.

[0029] Optionally, the determining each target event associated with the target information includes:

[0030] Determine each target event involved in the target information from the knowledge graph;

[0031] Alternatively, each target event involved in the target information is determined from a pre-created event table, wherein the event table stores a correspondence between information identifiers of each stored information and events.

[0032] Optionally, the method further comprises:

[0033] Grouping the third information entities and the extended entities to obtain multiple entity combinations, wherein the entity combinations include multiple information entities, and the information entities included in every two entity combinations are different;

[0034] Inputting the plurality of groups of entity combinations into an event reasoning model to obtain events associated with the plurality of groups of entity combinations;

[0035] A display mode of events associated with the plurality of entity combinations is determined and provided to the display interface. The event reasoning model is trained based on sample entity combinations and sample events associated with the sample entity combinations.

[0036] Optionally, determining the display method of each information entity, the information corresponding to each information entity, and the association relationship between each information entity according to a predetermined rule includes:

[0037] The display method of each information entity, the information corresponding to each information entity, and the relationship between each information entity is determined according to the structure of the knowledge graph.

[0038] Optionally, the method further comprises:

[0039] Reference information corresponding to the first query text is determined from a pre-constructed first reference information library, and the reference information associated with the query text is provided to the display interface, where the reference information is used to provide knowledge reference for users to edit manuscripts.

[0040] Optionally, the method further comprises:

[0041] Obtain update information, and update the knowledge graph based on the update information.

[0042] Optionally, the knowledge graph includes a plurality of information nodes and a plurality of entity nodes, the information nodes store stored information, and the entity nodes store information entities corresponding to the stored information;

[0043] The information entity corresponding to the stored information includes the event corresponding to the stored information;

[0044] There is a first connecting edge between every two information nodes corresponding to the same event in the knowledge graph, and the first connecting edge stores events corresponding to the information nodes connected to the connecting edge.

[0045] Optionally, the multiple entity nodes include multiple event nodes, and the event nodes store events corresponding to the stored information. There is a second connecting edge between every two event nodes with an associated relationship, and the second connecting edge stores the association relationship between two events corresponding to the two connected event nodes.

[0046] Optionally, the multiple entity nodes further include multiple character nodes, the character nodes store character information corresponding to the stored information, a third connection edge exists between every two character nodes having an associated relationship, the third connection edge stores the association relationship between two characters corresponding to the two connected character nodes, a fourth connection edge exists between the character node and an event node or an information node associated with the character node, the fourth connection edge stores the association relationship between the connected character node and the information node or the event node;

[0047] And / or, the multiple entity nodes also include multiple location nodes, the location nodes store the occurrence location information corresponding to the stored information, there is a fifth connection edge between every two location nodes with an associated relationship, the fifth connection edge stores the association relationship between the two locations corresponding to the two connected location nodes, there is a sixth connection edge between the location node and the event node or information node associated with the location node, the sixth connection edge stores the association relationship between the connected location node and the information node or event node.

[0048] Optionally, the process of constructing the knowledge graph includes:

[0049] Obtain each piece of information to be stored in the knowledge graph to be created;

[0050] Aggregating the information to be stored to obtain multiple information groups, wherein each information in each information group is similar information;

[0051] Determine each information entity corresponding to the information group according to the content of each information in the information group;

[0052] The knowledge graph is created based on the information entities corresponding to the information groups and the information to be stored.

[0053] Optionally, the information to be stored is aggregated to obtain multiple groups of information groups, including:

[0054] The information to be stored is aggregated based on an aggregation rule to obtain multiple information groups, wherein the aggregation rule is: the content similarity between each information in each information group is greater than a first preset threshold, and the occurrence time interval is less than a preset time interval.

[0055] In a second aspect, the embodiment of the present application further provides a search method based on a knowledge graph, which is applied to a client, and the method includes:

[0056] Acquire a first query text input by a user, where the first query text is used for searching;

[0057] Obtaining a display method of each information entity, information corresponding to each information entity, and association relationships between each information entity, and displaying them on a display interface;

[0058] Among them, each information entity includes a first information entity and a second information entity, the first information entity is associated with the query text, the first information entity represents the information entity involved in the information associated with the query text, the second information entity is an information entity screened out from a pre-constructed knowledge graph and associated with the first information entity, and the information corresponding to each information entity and the association relationship between each information entity are respectively determined from the knowledge graph.

[0059] Optionally, the method further comprises:

[0060] Reference information associated with the first query text is displayed on a display interface. The reference information is information associated with the first query text determined from a pre-constructed first reference information library. The reference information is used to provide knowledge reference for users to edit manuscripts.

[0061] In a third aspect, the embodiment of the present application further provides a search device based on a knowledge graph, which is applied to a server, and the device includes:

[0062] A first acquisition unit, used to acquire a first query text;

[0063] a determining unit, configured to determine a first information entity associated with the first query text, wherein the first information entity represents an information entity involved in the information associated with the query text;

[0064] A search unit, configured to filter out at least one second information entity associated with the first information entity from a pre-constructed knowledge graph; determine from the knowledge graph the information corresponding to each information entity and the association relationship between each information entity, wherein each information entity includes each first information entity and each second information entity;

[0065] The determining unit is further used to determine the display mode of each information entity, the information corresponding to each information entity, and the association relationship between each information entity according to a predetermined rule, and provide them to a display interface that needs to be displayed.

[0066] Optionally, the determination unit is specifically configured to: perform entity recognition on the first query text to obtain at least one third information entity; and determine a first information entity associated with the first query text according to each of the third information entities.

[0067] Optionally, the determination unit is specifically used to: perform entity expansion on the third information entity based on a pre-constructed knowledge graph to obtain at least one extended entity associated with the third information entity; determine a first information entity associated with the first query text according to the third information entity and the extended entity.

[0068] Optionally, the acquisition unit is further used to: acquire a second query text;

[0069] The determination unit is specifically used to: search for target information matching the second query text from the information stored in the pre-constructed knowledge graph; determine the information entities associated with the target information from the knowledge graph; and perform entity expansion on the third information entity based on the information entities associated with the target information; wherein the first query text is the text expanded on the basis of the second query text.

[0070] Optionally, the knowledge graph stores stored information and events, locations and people corresponding to the stored information;

[0071] The search unit is specifically used to: determine each target event involved in the target information; search for each target occurrence location and each target person associated with the target event from the knowledge graph; and determine the target event, the target occurrence location and the target person as information entities associated with the target information.

[0072] Optionally, the search unit is specifically used to: determine each target event involved in the target information from the knowledge graph; or, determine each target event involved in the target information from a pre-created event table, wherein the event table stores the correspondence between the information identifier of each stored information and the event.

[0073] Optionally, the device further comprises:

[0074] A grouping unit, used for grouping each of the third information entities and each of the extended entities to obtain a plurality of entity combinations, wherein the entity combinations include a plurality of information entities, and the information entities included in every two of the entity combinations are different;

[0075] An event reasoning unit, used for inputting the multiple groups of entity combinations into an event reasoning model to obtain events associated with the multiple groups of entity combinations;

[0076] The determining unit is further used to determine a display mode of events associated with the multiple groups of entity combinations, and provide the display mode to the display interface. The event reasoning model is trained based on sample entity combinations and sample events associated with the sample entity combinations.

[0077] Optionally, the determination unit is specifically used to determine the display method of each information entity, the information corresponding to each information entity, and the association relationship between the information entities according to the structure of the knowledge graph.

[0078] Optionally, the device further comprises:

[0079] The reference unit is used to determine the reference information corresponding to the first query text from a pre-constructed first reference information library, and provide the reference information associated with the query text to the display interface, wherein the reference information is used to provide knowledge reference for users to edit manuscripts.

[0080] Optionally, the device further comprises:

[0081] An updating unit is used to obtain update information and update the knowledge graph based on the update information.

[0082] Optionally, the knowledge graph includes multiple information nodes and multiple entity nodes, the information nodes store stored information, and the entity nodes store information entities corresponding to the stored information; the information entities corresponding to the stored information include events corresponding to the stored information; there is a first connecting edge between every two information nodes corresponding to the same event in the knowledge graph, and the first connecting edge stores events corresponding to the information nodes connected to the connecting edge.

[0083] Optionally, the multiple entity nodes include multiple event nodes, and the event nodes store events corresponding to the stored information. There is a second connecting edge between every two event nodes with an associated relationship, and the second connecting edge stores the association relationship between two events corresponding to the two connected event nodes.

[0084] Optionally, the multiple entity nodes further include multiple character nodes, the character nodes store character information corresponding to the stored information, a third connection edge exists between every two character nodes having an associated relationship, the third connection edge stores the association relationship between two characters corresponding to the two connected character nodes, a fourth connection edge exists between the character node and an event node or an information node associated with the character node, the fourth connection edge stores the association relationship between the connected character node and the information node or the event node;

[0085] And / or, the multiple entity nodes also include multiple location nodes, the location nodes store the occurrence location information corresponding to the stored information, there is a fifth connection edge between every two location nodes with an associated relationship, the fifth connection edge stores the association relationship between the two locations corresponding to the two connected location nodes, there is a sixth connection edge between the location node and the event node or information node associated with the location node, the sixth connection edge stores the association relationship between the connected location node and the information node or event node.

[0086] Optionally, the device further comprises:

[0087] A graph construction unit is used to obtain each information to be stored in the knowledge graph to be created; aggregate the information to be stored to obtain multiple groups of information groups, wherein each information in each information group is similar information; determine each information entity corresponding to the information group according to the content of each information in the information group; and create the knowledge graph based on each information entity corresponding to each information group and each information to be stored.

[0088] Optionally, the graph construction unit is specifically used to: aggregate each of the information to be stored based on an aggregation rule to obtain multiple groups of information groups, wherein the aggregation rule is: the content similarity between each information in each group of the information groups is greater than a first preset threshold, and the occurrence time interval is less than a preset time interval.

[0089] In a fourth aspect, an embodiment of the present application further provides a search device based on a knowledge graph, which is applied to a client, and the device includes:

[0090] A second acquisition unit, acquiring a first query text input by a user, wherein the first query text is used for searching;

[0091] A display unit, used to obtain the display mode of each information entity, the information corresponding to each information entity, and the association relationship between the information entities, and display them on a display interface;

[0092] Among them, each information entity includes a first information entity and a second information entity, the first information entity is associated with the query text, the first information entity represents the information entity involved in the information associated with the query text, the second information entity is an information entity screened out from a pre-constructed knowledge graph and associated with the first information entity, and the information corresponding to each information entity and the association relationship between each information entity are respectively determined from the knowledge graph.

[0093] Optionally, the display unit is further used for:

[0094] Reference information associated with the query text is displayed on a display interface. The reference information is information associated with the first query text determined from a pre-constructed first reference information library. The reference information is used to provide knowledge reference for users to edit manuscripts.

[0095] In a fifth aspect, an embodiment of the present application further provides an electronic device, including:

[0096] Processor; and

[0097] The memory is used to store a data processing program. After the electronic device is powered on and the program is run by the processor, the method described in any one of the first aspect or the second aspect is executed.

[0098] In a sixth aspect, an embodiment of the present application further provides a computer-readable storage medium storing a data processing program, which is executed by a processor to perform a method as described in any one of the first aspect or the second aspect.

[0099] Compared with the prior art, this application has the following advantages:

[0100] The knowledge graph-based search method provided in the present application determines the first information entity involved in the information corresponding to the first query text, and then determines the information corresponding to each information entity and the relationship between each information entity from the pre-constructed knowledge graph. Since the search in the present application is based on the knowledge graph and the properties of the knowledge graph itself, that is, there is an edge between two nodes with an association relationship in the knowledge graph, and the edge can represent the relationship between the contents corresponding to the two nodes connected to it, the second information entity associated with the first information entity and the relationship between the first information entity and the information entities contained in the second information entity can be quickly and accurately found through the pre-constructed knowledge graph, and the information corresponding to each information entity can be easily found based on the knowledge graph.

[0101] After the second information entity associated with the first information entity, the information corresponding to each information entity, the relationship between each information entity, and the display method of these contents are determined, the determined contents can be displayed on the display interface according to the display method, and planners can view the determined contents on the display interface. Since the displayed contents include each information entity, the information corresponding to the information entities, and the relationship between the information entities, the displayed contents are the contents that have been sorted out and searched for relationships for planners. Planners can conveniently, clearly and comprehensively view the information related to the first query text entered by themselves through the displayed contents. Since the displayed information is more comprehensive and the relationship between each information has been sorted out, planners do not need to conduct multiple searches repeatedly, nor do they need to sort out the searched contents too much, which simplifies the information search process before planners edit the information, thereby improving the efficiency of planners in editing information.

[0102] In addition, the embodiment of the present application determines relevant information and entities from the knowledge graph based on each first information entity associated with the query text. Since the first information entity can not only express the search intent of the first query text more accurately, but also make information search faster and more efficient due to its high information efficiency, compared to searching directly through the query text, the first information entity is the information obtained after filtering out non-critical information in the query text. Therefore, information search can be performed efficiently through the first information entity.

[0103] In an optional solution, the determining of each target event associated with the target information includes: determining each target event involved in the target information from the knowledge graph. In this way, each target event involved in the target information can be easily determined through the knowledge graph.

[0104] In another optional solution, the determining each target event associated with the target information includes: determining each target event involved in the target information from a pre-created event table, wherein the event table stores the corresponding relationship between the information identifier of each stored information and the event. In this way, each target information event involved in the target information can be quickly determined through the event table, thereby improving the efficiency of information search.

[0105] In an optional scheme, the multiple entity nodes also include multiple character nodes, and the character nodes store character information corresponding to the stored information. There is a third connecting edge between every two character nodes with an associated relationship, and the third connecting edge stores the association relationship between the two characters corresponding to the two connected character nodes. There is a fourth connecting edge between the character node and the event node or information node associated with the character node, and the fourth connecting edge stores the association relationship between the connected character node and the information node or event node.

[0106] By setting character nodes and setting the third connection edge and the fourth connection edge, this solution can search out and display characters related to information, characters related to events, and characters with associated relationships during the search, so that planners can know the associations between information, events, and characters, making it easier for planners to plan.

[0107] In an optional scheme, the multiple entity nodes also include multiple location nodes, and the location nodes store the occurrence location information corresponding to the stored information. There is a fifth connecting edge between every two location nodes with an associated relationship, and the fifth connecting edge stores the association relationship between the two locations corresponding to the two connected location nodes. There is a sixth connecting edge between the location node and the event node or information node associated with the location node, and the sixth connecting edge stores the association relationship between the connected location node and the information node or event node.

[0108] By setting the location node and the fifth and sixth connection edges, this solution can search out and display the locations related to information, locations related to events, and locations with associated relationships during the search, so that planners can know the associated relationships between information, events, and locations, making it easier for planners to carry out news planning. BRIEF DESCRIPTION OF THE DRAWINGS

[0109] Figure 1 is a flowchart of an example of a search method provided in an embodiment of the present application;

[0110] Figure 2 yes Figure 1A flowchart of an example of determining a first information entity in the search method shown;

[0111] Figure 3 yes Figure 2 A flowchart of another example of determining a first information entity in the method shown;

[0112] Figure 4 Shown is a schematic diagram of a knowledge graph provided in an embodiment of the present application;

[0113] Figure 5 is a flowchart of another example of the search method provided in the embodiment of the present application;

[0114] Figure 6 is a unit block diagram of a search device provided in an embodiment of the present application;

[0115] Figure 7 It is a structural diagram of an electronic device for implementing a search method provided in an embodiment of the present application.

[0116] Figure 8 It is a structural diagram of the search system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0117] Many specific details are described in the following description to facilitate a full understanding of the present application. However, the present application can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the connotation of the present application, so the present application is not limited by the specific implementation disclosed below.

[0118] After nearly a decade of rapid development of the Internet industry, the growth rate and richness of information are far beyond the past. The emergence of self-media, Weibo, etc. and the need for transformation and upgrading of traditional media have further accelerated the process of information growth. The speed of information generation can gradually reach all aspects of people's lives in seconds. However, with the popularization and popularization of information, the entry threshold for information practitioners has gradually decreased, and the emergence of more low-quality or even factually incorrect and out-of-context information has gradually become a problem in the industry. For professional information practitioners (such as information editors, information reporters, etc.), with the gradual increase in the perceptible sensitivity of information events and the gradual enrichment of information event details, empirical planning has gradually failed to meet readers' reading needs for information. Editors need to invest more time and energy to sort out the required content from the vast, real-time incremental, and dynamically changing multimodal information base data containing images, texts, and videos, in order to form appropriate, complete, and correct information manuscripts. Such tedious work seriously squeezes the time for information practitioners to think deeply, affecting the quality of information. Therefore, the planning industry urgently needs a set of search solutions to address the pain points of newspaper reporters, editors, planners and other personnel and improve work efficiency.

[0119] Based on the above problems, the embodiments of the present application provide a search method, device, electronic device and computer-readable storage medium based on a knowledge graph, which can quickly search and sort out the information required by planners, thereby improving the efficiency of planners in editing information.

[0120] The first embodiment of the present application provides a search method based on a knowledge graph, and the executor of the method is a server. Specifically, the executor of the method is a server device, which can be a server, a desktop computer, a laptop computer, a smart mobile terminal, etc., or other electronic devices that can serve as a server.

[0121] The knowledge graph-based search method provided in the embodiment of the present application can be used as an auxiliary means for planners or information editors to edit information. It can quickly search and sort out the information required by planners, thereby improving the efficiency of planners in editing information.

[0122] like Figure 1 As shown, the knowledge graph-based search method provided in the first embodiment of the present application includes the following steps S110 to S150.

[0123] Step S110: Acquire a first query text.

[0124] The first query text is the first query text input by the planner when searching, and is used to search for information matching the first query text.

[0125] In step S110 , the planner may input a first query text on the client, and the client sends the first query text input by the user to the server, and the server receives the first query text sent from the client, so that the server can obtain the first query text.

[0126] The first query text may be a Chinese text, or a foreign language text such as English, French, German, etc., which is not specifically limited in this application.

[0127] Step S120: Determine at least one first information entity associated with the first query text.

[0128] The first information entity is used to represent the information entity involved in the information corresponding to the first query text.

[0129] The information corresponding to the first query text may include information corresponding to the first query text, papers, books, legal clauses, etc., but is not limited thereto. For ease of description, the following mainly introduces the present application with the information corresponding to the first query text including the information corresponding to the first query text.

[0130] The information entities involved in the information can also be understood as the named entities involved in the information. Named entities are entities identified by names. Named entities may include at least one of: people, events, places, times, organizations, product models, sports events, and solar terms. They may also include other named entities involved in the information, which are not specifically limited in this application.

[0131] In step S120, the first information entity associated with the first query text may be determined based on entity recognition. Specifically, entity recognition may be performed on the first query text, and the recognized entity may be used as the first information entity associated with the first query text.

[0132] Specifically, entity recognition can be performed on the first query text based on the NER algorithm to obtain recognized entities. The NER algorithm is also called proper name recognition, and the NER algorithm can extract entities such as names, place names, and proper nouns from unstructured input text. Alternatively, entity recognition can also be performed based on other entity recognition algorithms, which are not specifically limited in this application.

[0133] The first information entity determined in step S120 may include one entity or multiple entities. When the first information entity includes multiple entities, the scope of subsequent search may be more precise, thereby increasing the accuracy of the search.

[0134] Step S130: Filter out at least one second information entity associated with the first information entity from the pre-constructed knowledge graph.

[0135] Step S140: Determine from the knowledge graph the information corresponding to each information entity and the relationship between each information entity.

[0136] The above-mentioned information entities include first information entities and second information entities.

[0137] The knowledge graph can be understood as a semantic network that can describe objective things in the form of a graph. The graph here contains nodes and edges. Nodes represent entities, and edges represent the relationship between entities. The entity can be a person, a place, an organization, a concept, an article, etc. In the knowledge graph, a node-edge-node can be regarded as a record, a record is a triple data, and the knowledge graph contains multiple triple data.

[0138] It can be understood that the knowledge graph pre-constructed in the embodiment of the present application is used for information search. The knowledge graph contains multiple entity nodes and multiple information nodes. The two entity nodes corresponding to two information entities with an association relationship are connected by a relationship edge, the information entity with an association relationship and the information are connected by a relationship edge, and the two information nodes with an association relationship are connected by a relationship edge.

[0139] Taking information search as an example, the knowledge graph includes multiple information nodes and multiple entity nodes. Information is stored on the information node, and entities corresponding to the information are stored on the entity node. In the embodiment of the present application, one information node of the knowledge graph can store one information. Alternatively, multiple information can also be stored on an information node. When multiple information is stored on an information node, the multiple information is usually similar information, that is, information with high similarity. In this way, the structure of the knowledge graph can be made simpler by clustering, so that the search speed can be faster. The information stored on the information node may include information such as the text of the information, information report video, information title, information occurrence date, and the link address (such as URL) corresponding to the information, but is not limited to this.

[0140] In steps S130 to S140, a first entity node that is associated with an entity node corresponding to a first information entity can be determined from the knowledge graph, and the information entity corresponding to the first entity node can be determined as a second information entity associated with the first information entity; a first information node that is associated with an entity node corresponding to the first information entity can be determined from the knowledge graph, and information corresponding to the first information node can be determined as information corresponding to the first information entity; second entity nodes corresponding to each information entity can be determined from the knowledge graph, and association relationships between each information entity corresponding to each second entity node can be determined based on the connecting edges between each second entity node.

[0141] In the embodiment of the present application, based on the knowledge graph, the second information entity associated with the first information entity, the information corresponding to the first information entity, and the association relationship between the information entities can be easily determined.

[0142] Step S150: Determine the display method of each information entity, the information corresponding to each information entity, and the association relationship between each information entity according to a predetermined rule, and provide them to the display interface that needs to be displayed.

[0143] In step S150, the server may send each information entity, the information corresponding to each information entity, the association between each information entity, and the above-mentioned display method relationship to the client. After receiving this information, the client may display the information on the display interface according to the display method, so that planners can view the information through the display interface of the client.

[0144] The above-mentioned predetermined rule may be to display the information entities, the information corresponding to each information entity, and the association relationship between each information entity in a table, a graphical form or other display forms, which is not specifically limited in this application.

[0145] The above-mentioned information entity may include at least one of the event, place of occurrence, person, time, and proper name corresponding to the information, but is not limited thereto.

[0146] The knowledge graph-based search method provided in the present application determines the first information entity involved in the information corresponding to the first query text, and then determines the information corresponding to each information entity and the relationship between each information entity from the pre-constructed knowledge graph. Since the search in the present application is based on the knowledge graph, the knowledge graph usually contains various nodes, and each node is connected by a relationship edge. The knowledge graph stores various information, the entities corresponding to each information, and the relationship between each information and each entity through each node and relationship edge.

[0147] Based on the properties of the knowledge graph itself, that is, there is an edge between two nodes with an association relationship in the knowledge graph, and the edge can represent the relationship between the contents corresponding to the two nodes connected to it, the pre-constructed knowledge graph can quickly and accurately find the second information entity associated with the first information entity, the association relationship between the first information entity and the information entities contained in the second information entity, and the information corresponding to each information entity can be easily found based on the knowledge graph.

[0148] After the second information entity associated with the first information entity, the information corresponding to each information entity, the relationship between each information entity, and the display method of these contents are determined, the determined contents can be displayed on the display interface according to the display method, and planners can view the determined contents on the display interface. Since the displayed contents include each information entity, the information corresponding to the information entities, and the relationship between the information entities, the displayed contents are the contents that have been sorted out and searched for relationships for planners. Planners can conveniently, clearly and comprehensively view the information related to the first query text entered by themselves through the displayed contents. Since the displayed information is more comprehensive and the relationship between each information has been sorted out, planners do not need to conduct multiple searches repeatedly, nor do they need to sort out the searched contents too much, which simplifies the information search process before planners edit the information, thereby improving the efficiency of planners in editing information.

[0149] In addition, the embodiment of the present application determines relevant information and entities from the knowledge graph based on each first information entity associated with the query text. Since the first information entity can not only express the search intent of the first query text more accurately, but also make information search faster and more efficient due to its high information efficiency, compared to searching directly through the query text, the first information entity is the information obtained after filtering out non-critical information in the query text. Therefore, information search can be performed efficiently through the first information entity.

[0150] In one embodiment, Figure 2 As shown, step S120 can be implemented according to the following steps S121 to S122.

[0151] Step S121: Perform entity recognition on the first query text to obtain at least one third information entity.

[0152] Step S122: Determine the first information entity associated with the first query text according to each third information entity.

[0153] The method of performing entity recognition in step S121 can refer to the process of using the NER algorithm to perform entity recognition in step S120 above, and the specific method will not be described in detail.

[0154] In step S122, the third information entity may be directly determined as the first information entity associated with the first query text.

[0155] This implementation manner can conveniently, quickly and accurately obtain the first information entity associated with the first query text by performing entity recognition on the first query text.

[0156] In a specific embodiment, Figure 3 As shown, step S122 can be implemented according to the following steps S122a-S122b.

[0157] Step S122a: Perform entity expansion on the third information entity based on the pre-constructed knowledge graph to obtain at least one extended entity associated with the third information entity.

[0158] Step S122b: Determine the first information entity associated with the first query text according to the third information entity and the extended entity.

[0159] In step S122a, an information entity having an association relationship with the third information entity can be determined from a pre-constructed knowledge graph, and the information entity having an association relationship with the third information entity can be determined as the above-mentioned extended entity.

[0160] For example, the third information entity is "Sports Games A", and the information entities associated with "Sports Games A" in the knowledge graph include "Location A", "Sports A", "Ice and Snow", and "Host". These information entities associated with "Sports Games A" are extended entities of "Sports Games A".

[0161] In step S122b, each third information entity and each extended entity may be determined as a first information entity associated with the first query text, or each third information entity, each extended entity and other information entities associated with the first query text may be determined as a first information entity.

[0162] This embodiment performs entity expansion on the third information entity based on the knowledge graph, and determines the first information entity based on the obtained expanded entity and the third information entity. In this way, the determined first information entity is a plurality of information entities after expansion. The determined second information entity, the information corresponding to the first information entity, and each information entity can be covered more widely based on the plurality of information entities after expansion, so that the scope of the searched information is also wider, so that the information and information entities associated with the first query text can be searched more comprehensively, reducing the probability of planners performing multiple searches, thereby further improving the efficiency of planners in planning.

[0163] In a specific embodiment, the above method may further include the following steps S140 to S150.

[0164] Step S140: grouping the third information entities and the extended entities to obtain multiple entity combinations.

[0165] Each entity combination contains multiple information entities, and there are differences between the information entities contained in each two entity combinations. The multiple information entities in each entity combination can usually describe one or more events, which can be, for example, information events, academic events, invention events, etc., but are not limited thereto.

[0166] There are differences in the information entities contained in each combination of two groups of entities. The information entities contained in the two groups of entity combinations may be different, that is, there are no identical information entities in the two groups of entity combinations. Alternatively, there may be identical information entities and different information entities in the two groups of entity combinations, that is, there are differences in the information entity parts of the two groups of information entity combinations.

[0167] For example, the entity combination [Location B, Host, Athlete] can describe two information events: Sports Meeting B held at Location B and Sports Meeting A held at Location B. The entity combination [Ice and Snow, Location B, Host] can describe the information event: Sports Meeting A held at Location. The entity combination [Sports Meeting A, Solar Term A] can describe the information event: Sports Meeting A held at Location B. The entity combination [Location B, Summer, Athlete] can describe the information event: Sports Meeting B held at Location B.

[0168] In the process of grouping, each third information entity and each information entity in each extended entity can be randomly grouped for combination; or, grouping can be performed according to the principle that the entity types of each information entity in the same entity combination are different. For example, the types of the three information entities in the entity combination [Location B, Summer, Athlete] are location, season, and person, respectively. This application does not limit the specific method of grouping.

[0169] Step S150: input multiple entity combinations into the event reasoning model to obtain information events associated with the multiple entity combinations, and determine the display method of the events associated with the multiple entity combinations to provide to the display interface.

[0170] The above event reasoning model is obtained by training based on sample entity combinations and sample events associated with the sample entity combinations. For example, the event reasoning model can be obtained by training a model to be trained through sample entity combinations and sample events. The model to be trained can be a neural network model, a logistic regression model, etc., or other deep learning models.

[0171] For example, in step S150, the server may send information events associated with multiple groups of entity combinations and the display method of each associated information event to the client. After receiving the information events associated with the multiple groups of entity combinations, the client may display the received information events in the display method on the display interface, so that planners can view the information events associated with the multiple groups of entity combinations through the display interface of the client. In this way, planners can know which information events the information they want to search for involves, which provides planners with a clearer sorting of information events, makes it easier for planners to plan, and improves the information editing efficiency of planners.

[0172] This embodiment obtains multiple entity combinations by grouping various information entities. Since the entity combination contains at least multiple information entities, multiple information entities can more accurately describe an information event than one information entity. Therefore, the information event obtained by entity combination is usually more accurate, thereby providing planners with more accurate information event information, reducing the event screening process for planners and improving planning efficiency.

[0173] After determining the information events associated with the multiple entity combinations in step S150, the information events associated with the multiple entity combinations can also be determined as the first information entity. In this way, the determined first information entity can be more comprehensive, thereby making the searched information and information entity more comprehensive and accurate.

[0174] In a specific embodiment, before step S110, the following steps S1 to S3 may also be included.

[0175] Step S1: Obtain a second query text.

[0176] The second query text is obtained in a similar manner to the first query text, and will not be described in detail here.

[0177] Step S2: Search for target information matching the second query text from the information stored in the pre-constructed knowledge graph.

[0178] In step S2, taking the information search and the information stored in the knowledge graph as information as an example, a search can be performed based on text similarity, and information similar to the first query text can be searched from the information stored in the knowledge graph as target information. The target information is the target information. In this case, the target information matching the second query text can also be understood as similar information of the second query text. Specifically, the information containing the second query text can be determined as the target information, or the information containing the first text can be determined as the target information. The text similarity between the first text and the second query text is greater than a preset threshold, and the preset threshold can be, for example, 70% to 90%. For example, the preset threshold can be 70%, 80%, 90%, or a threshold within another range.

[0179] Step S2 may also search for information matching the second query text based on other methods, which will not be described in detail here.

[0180] Step S3: Determine each information entity associated with the target information from the knowledge graph.

[0181] In step S3, each information entity corresponding to each entity node connected to the information node corresponding to the target information may be determined as each information entity associated with the target information.

[0182] In step S122a, entity expansion may be performed according to the following step S4.

[0183] Step S4: Entity expansion is performed on the third information entity based on each information entity associated with the target information.

[0184] In step S4, each information entity associated with the target information may be determined as an extended entity.

[0185] This embodiment first finds the target information that matches the first query text, and then performs entity expansion through each information entity associated with the target information. Since the target information that matches the first query text can more comprehensively reflect the information that the planner wants to search for, the each information entity associated with the target information can also more comprehensively reflect the information that the planner wants to search for. Based on the each information entity associated with the target information, the third information entity is expanded in entity, so that each expanded entity obtained can be more comprehensive, thereby making the search result more comprehensive, and further improving the efficiency of the planner in planning.

[0186] In this embodiment, steps S1 to S3 can be understood as a preliminary search. The target information determined by the preliminary search and the information entities associated with the target information can be understood as a sub-graph that matches the search intent screened out from the knowledge graph. In this way, the subsequent search based on the first query text can have a smaller search scope and higher search precision and accuracy.

[0187] In one implementation, the above search method may further include the following steps:

[0188] Determine a display mode of target information and each information entity associated with the target information, and provide the display mode to a display interface, so as to display the target information and each information entity associated with the target information on the display interface through the display mode.

[0189] Specifically, the server can send the target information, each information entity associated with the target information, and the display method to the client, and the client can display the target information, each information entity associated with the target information, and the relationship between each information entity on the display interface according to the display method. In this way, the planner can expand the second query text according to the displayed content to obtain a more detailed first query text, thereby making the search more accurate.

[0190] In a specific embodiment, the above-mentioned knowledge graph stores stored information and events, places of occurrence, and people corresponding to the stored information.

[0191] Step S3 can be implemented by following steps S3a to S3c.

[0192] Step S3a: Determine each target event involved in the target information.

[0193] In step S3a, each target event involved in the target information can be determined from the knowledge graph. Specifically, the entity nodes in the knowledge graph may include event nodes storing events, location nodes storing location nodes, and character nodes storing character information. The server can determine the events corresponding to the event nodes that have a connection relationship with the information nodes corresponding to the target information as the target events involved in the target information. This embodiment can easily determine the target events involved in the target information through the knowledge graph.

[0194] Figure 4 FIG. 1 is a schematic diagram of a knowledge graph provided in an embodiment of the present application, for example, Figure 4 As shown, the "Place B" node and the "Place A" node are place nodes, the "Sports Meeting A" node and the "Sports Event A" node are event nodes, and the "Director Zhang" node and the "Actor" node are character nodes.

[0195] In step S3a, each target event involved in the target information can also be determined from a pre-created event table, and the event table stores the corresponding relationship between the information identifier of each stored information and the event. Taking the above information search as an example, the stored information is the stored information, and the information identifier of the stored information can be the title, summary or text of the stored information, or other information identifiers, which are not specifically limited here. This embodiment can quickly determine each target information event involved in the target information through the event table, thereby improving the efficiency of information search.

[0196] Step S3b: Search the knowledge graph for each target location and each target person associated with the target event.

[0197] In step S3b, the location node and person node that are connected to the event node corresponding to the target event in the knowledge graph respectively correspond to the location and person as the target location and target person.

[0198] Step S3c: Determine the target event, target location and target person as information entities associated with the target information.

[0199] In this embodiment, since an event is an entity that describes information more accurately, information can be summarized very accurately through an event, so the target event involved in the target information can summarize the target information more accurately, and the target location and target person can be determined more quickly and accurately based on the target event, and the determined target location and target person are also more comprehensive.

[0200] In one implementation, the above search method may further include the following step S160.

[0201] Step S160: Determine reference information corresponding to the first query text from a pre-constructed first reference information library, and provide the reference information associated with the first query text to a display interface for display.

[0202] The above-mentioned reference information is used to provide knowledge reference for users to edit manuscripts. The above-mentioned reference information may include speech content, fixed sentence patterns, guiding documents / spirits, legal terms, scientific knowledge, writing templates, etc., but is not limited thereto.

[0203] For example, the guiding document / spirit can be: the officially issued "Medium- and Long-Term Youth Development Plan (2016-2025)", the content of the person's speech: the speech of a specific person, etc., but not limited to these.

[0204] The first reference information database may store reference information and information entities corresponding to the reference information. The information entity corresponding to the reference information may be at least one of an event, a place, and a person. In step S160, the information entity corresponding to the first query text may be determined first, and then the reference information corresponding to the information entity corresponding to the first query text may be determined from the first reference information database. The process of determining the information entity corresponding to the first query text may refer to the detailed description of step S120 and step S3a above, which will not be repeated here.

[0205] In this implementation, the server may send the reference information corresponding to the first query text to the client, and the client may display the reference information associated with the first query text on a display interface.

[0206] By determining reference information, this implementation mode allows planners to edit information according to the reference information, for example, by applying fixed sentence patterns, quoting speech content of characters, etc., so that planners can achieve efficient information editing.

[0207] In one implementation, the above method may further include the following step S170.

[0208] Step S170: Obtain update information, and update the above knowledge graph based on the update information.

[0209] The above-mentioned update information is information not stored in the knowledge graph. In step S170, the update information may be obtained from other information storage devices, or uploaded by background maintenance personnel, or obtained by other means.

[0210] In an embodiment of the present application, the knowledge graph has an automated incremental link, and the process of updating the knowledge graph based on the updated information can be implemented by configuring config, that is, by configuring config, it is possible to align the updated information with the pre-created knowledge graph.

[0211] When the updated information is aligned to the knowledge graph, corresponding information nodes and entity nodes will be added to the knowledge graph to store the updated information and the information entities corresponding to the updated information.

[0212] This implementation can realize the privatization construction of private domain data. By updating the knowledge graph, the information storage of the knowledge graph is more comprehensive, thereby making the search more accurate.

[0213] In one embodiment, the information entity corresponding to the above-mentioned stored information includes an event corresponding to the stored information; there is a first connecting edge between every two information nodes corresponding to the same event in the above-mentioned knowledge graph, and the first connecting edge stores events corresponding to the information nodes connected to the connecting edge.

[0214] That is, an edge is built between every two information nodes corresponding to the same event, and the built edge is used to represent the events corresponding to the two information nodes.

[0215] like Figure 4 As shown, taking information search as an example, various information are stored in the knowledge graph, and an edge can be built between the two information nodes "Sports Meeting A Information 1" and "Sports Meeting A Information 3", which can indicate that both are information about Sports Meeting A.

[0216] This implementation uses an event as an edge to connect two information nodes, which can easily establish an association relationship between the two information nodes, and can also easily find the information corresponding to the event and the event corresponding to the information during the search, thereby improving the search efficiency.

[0217] In a specific embodiment, multiple entity nodes in the knowledge graph include multiple event nodes, and events corresponding to the stored information are stored on the event nodes. There is a second connecting edge between every two event nodes with an associated relationship, and the second connecting edge stores the association relationship between the two events corresponding to the two connected event nodes.

[0218] Every two event nodes with an associated relationship are two event nodes corresponding to every two events with an associated relationship.

[0219] For example, the association relationship between two information events may be: the association relationship between event "Sports Meeting A" and event "Sports Meeting C" is: a world-scale sports meeting.

[0220] For example, Figure 4 As shown, the edge between the "Sports Meeting A" event node and the "Sports Event A" event node can indicate that Sports Event A is a competition event of Sports Meeting A.

[0221] This embodiment establishes a relationship edge between two event nodes with an associated relationship, which can easily find various associated events during the search, thereby improving the relevance and coverage of the search, thereby improving the efficiency of the search.

[0222] In the embodiment of the present application, information nodes and event nodes can be regarded as the knowledge graph, which is used to search for information and information events. The knowledge graph contains two levels: [information, event, information] and [event, relationship between events, event]. Based on the two levels, information corresponding to the event and the associated events of the event can be efficiently and accurately searched, so that the search can be realized efficiently.

[0223] In a specific embodiment, the above-mentioned multiple entity nodes may also include multiple character nodes, and the character nodes store character information corresponding to the stored information. There is a third connection edge between every two character nodes with an associated relationship, and the third connection edge stores the association relationship between the two characters corresponding to the two connected character nodes. There is a fourth connection edge between the character node and the event node or information node associated with the character node, and the fourth connection edge stores the association relationship between the connected character node and the information node or event node.

[0224] For example, Figure 4 As shown, the connection edge between the character node "An" and the character node "Any Player" can indicate that An is the coach of the player. The connection edge between the character node "Any Player" and the event node "Sports A" can indicate that the player is a contestant of Sports A.

[0225] Specifically, the character information stored in the character node may include the character's name, resume, image, title, etc., but is not limited thereto.

[0226] By setting character nodes and setting the third connection edge and the fourth connection edge, this embodiment can search out and display characters related to information, characters related to events, and characters with associated relationships during the search, so that planners can know the associated relationships between information, events, and characters, which is more convenient for planners to plan.

[0227] In another specific embodiment, the above-mentioned multiple entity nodes may also include multiple location nodes, and the location nodes store the occurrence location information corresponding to the stored information. There is a fifth connection edge between every two location nodes with an associated relationship, and the fifth connection edge stores the association relationship between the two locations corresponding to the two connected location nodes. There is a sixth connection edge between the location node and the event node or information node associated with the location node, and the sixth connection edge stores the association relationship between the connected location node and the information node or event node.

[0228] The occurrence location information stored in the location node may include the name, image, longitude and latitude, regional division and other information of the occurrence location, but is not limited thereto.

[0229] By setting the location node and the fifth connection edge and the sixth connection edge, this embodiment can search out and display the locations related to the information, the locations related to the event, and the locations with associated relationships during the search, so that planners can know the associated relationships between the information, events, and locations, making it easier for planners to carry out news planning.

[0230] Optionally, there may be a seventh connection edge between the place node and the person node associated with the place node, and the sixth connection edge stores the association relationship between the connected place node and the person node.

[0231] In one implementation, the construction process of the above-mentioned knowledge graph may include the following steps S10 to S40.

[0232] Step S10: Obtain each piece of information to be stored in the knowledge graph to be created.

[0233] In step S10, the information to be stored can be obtained from other storage devices, or downloaded from the Internet, or the information to be stored can be manually input into the server, and the server obtains the manually input information to be stored. This application does not limit the specific method of obtaining the information to be stored.

[0234] The information to be stored may be news, papers, electronic books, online articles, etc. to be stored, but is not limited thereto.

[0235] Step S20: Aggregate each information to be stored to obtain multiple groups of information.

[0236] Among them, each information in each information group is similar information. Each information in each information group is similar information, and the similarity between the information content of each information in each information group may be greater than a first preset threshold. Specifically, the similarity between the information content of every two information in each information group may be greater than the first preset threshold; or, the content similarity between each information in each information group and the reference information may be greater than the first preset threshold, and the reference information may be any information in the information group. The first preset threshold may be, for example, 65% to 95%, for example, the first preset threshold may be 65%, 80%, 95%, or a threshold within another range.

[0237] In this implementation manner, whether two pieces of information are similar can be determined based on text similarity, which will not be described in detail here.

[0238] Step S30: Determine each information entity corresponding to the information group according to the content of each information in the information group.

[0239] In step S30, entity recognition can be performed on the content of each information in the information group to obtain each identified information entity, and each information entity corresponding to the information group can be determined based on each identified information entity. Specifically, each identified information entity can be determined as each information entity corresponding to the information group, or the target information entity among the identified information entities can be determined as each information entity corresponding to the information group. The target information entity is an information entity that each information in the information group has, that is, the target information entity can be identified in each information in the information group. Alternatively, each information entity corresponding to the information group can also be determined by other methods, which are not specifically limited in this application.

[0240] Step S40: Create a knowledge graph based on each information entity and each information to be stored corresponding to each information group.

[0241] In step S40, each information entity corresponding to each information group can be stored in each entity node, each information to be stored can be stored in the information node, and the connection edges can be set according to the relationship between each information entity and each information to be stored, so as to form the above-mentioned knowledge graph. The specific creation process of the knowledge graph will not be described in detail.

[0242] This implementation can conveniently summarize the information entities corresponding to each piece of information through information aggregation. Since the amount of information to be stored is usually very large, each information entity can be quickly determined through aggregation, thereby improving the efficiency of knowledge graph creation.

[0243] In a specific embodiment, step S20 may be implemented according to the following process: aggregating each of the information to be stored based on an aggregation rule to obtain multiple groups of information.

[0244] The above aggregation rule is: the content similarity between each information in each information group is greater than a first preset threshold, and the occurrence time interval is less than a preset time interval.

[0245] The time interval between each information in each information group is less than the preset time interval, which may be that the time interval between each two information in each information group is less than the preset time interval, or that the time interval between each information in each information group and the above-mentioned reference information is less than the preset time interval. The preset time interval may be, for example, 3 hours to 24 hours, for example, 3 hours, 10 hours, 24 hours, or a time interval within another range.

[0246] In one embodiment, the above-mentioned search method may further include the following steps: determining the information corresponding to the second information entity from the knowledge graph for displaying the information corresponding to the second information entity, so that planners can also view the information corresponding to the second information entity, making the search more comprehensive and further reducing the process of active search and summarization by planners.

[0247] In one implementation, before step S130, the search method may further include the following steps: determining high-frequency words from the first query text.

[0248] High-frequency words may be words of the same type in the first query text that appear more than a set number of times. For example, if the words "folk music", "orchestra", "music" and the like appear many times in the first query text, then words related to music are determined as high-frequency words.

[0249] Step S130 can be implemented according to the following process: determining information corresponding to the first information entity from a pre-constructed knowledge graph based on a first search principle.

[0250] The first search principle includes: the number of high-frequency words contained in the information is proportional to the probability of determining the information as the information corresponding to the first information entity. In other words, among the information corresponding to the first information entity in the knowledge graph, if a certain information contains a large number of high-frequency words, the probability of the information being determined as the information corresponding to the first information entity is higher.

[0251] This embodiment can make the determined information corresponding to the first information entity more consistent with the search intention of the planner.

[0252] In an optional embodiment, the present application further provides an information proofreading method, which is applied to a server, and the method includes:

[0253] Get the manuscript to be proofread;

[0254] Searching for target reference information from a pre-built second reference information database, wherein the target reference information is consistent with the text information cited in the manuscript to be proofread;

[0255] The text information cited in the manuscript to be proofread is proofread based on the target reference information to obtain a proofreading result, which is used to display the proofreading result on a display interface.

[0256] The above-mentioned manuscript to be proofread can be a manuscript written by a planner. The above-mentioned second reference information database can include the content of a person's speech, guiding documents / spirits, legal clauses, etc. The content in the second reference information database is usually information that does not allow any form of error.

[0257] In the embodiment of the present application, the target reference information can be searched from a pre-constructed second reference information library based on text similarity, and the specific method will not be described in detail here.

[0258] The target reference information is consistent with the text information cited in the manuscript to be proofread, and the similarity between the target reference information and the text information cited in the manuscript to be proofread is greater than a set threshold. The set threshold can be 80% to 95%, for example, 80%, 87%, 95%, or other larger percentage values, such as 75%, 98%, which is not specifically limited in this application.

[0259] When proofreading the text information cited in the manuscript to be proofread based on the target reference information, the text information cited in the manuscript to be proofread and the target reference information can be proofread word by word, and the proofreading results obtained may include: wrong characters, missing characters, extra characters, different positions, modification suggestions, etc., but are not limited thereto. Specifically, during word-by-word proofreading, feature descriptors can be generated word by word for the target reference information and the text information cited in the manuscript to be proofread, and the proofreading results can be obtained by comparing the feature descriptors corresponding to each word.

[0260] In this embodiment, the server can send the proofreading result to the client, and the client can display the proofreading result on the display interface after receiving the proofreading result. The server can also send the target reference information to the client, and the client displays the target reference information on the display interface, so that the planner can modify the edited manuscript according to the displayed proofreading result and target reference information.

[0261] The second embodiment of the present application also provides a search method based on a knowledge graph, and the executor of the method is a client. Specifically, the executor of the method is a client device, and the client device can be a desktop computer, a laptop computer, a smart mobile terminal, a mobile phone, a tablet computer, a smart watch, etc., or other electronic devices that can serve as a client.

[0262] like Figure 5 As shown, the knowledge graph-based search method provided in the second embodiment of the present application includes the following steps S210 to S220.

[0263] Step S210: Acquire a first query text input by a user.

[0264] The first query text is used to perform information search.

[0265] Step S220: Acquire each information entity, the information corresponding to each information entity, and the display method of the association relationship between each information entity, and display them on the display interface.

[0266] Among them, each information entity includes a first information entity and a second information entity. The first information entity is associated with the query text. The first information entity represents the information entity involved in the information associated with the query text. The second information entity is an information entity screened from a pre-constructed knowledge graph and associated with the first information entity. The information corresponding to each information entity and the association relationship between each information entity are determined from the knowledge graph respectively.

[0267] In one implementation, the knowledge graph-based search method provided in the second embodiment of the present application also includes the following step S240.

[0268] Step S240: displaying reference information associated with the first query text on the display interface. The reference information is information associated with the first query text determined from a pre-constructed first reference information library. The reference information is used to provide knowledge reference for users to edit manuscripts.

[0269] Since the search method provided in the second embodiment of the present application has corresponding specific technical features with the scheme of the search method provided in the first embodiment, the specific description and beneficial effects of the relevant features can be referred to the first embodiment and will not be repeated here.

[0270] Corresponding to the search method based on knowledge graph provided in the first embodiment of the present application, the third embodiment of the present application provides a search device based on knowledge graph, which is applied to the server, such as Figure 6 As shown, the device includes:

[0271] A first acquisition unit, used to acquire a first query text;

[0272] a determining unit, configured to determine a first information entity associated with the first query text, wherein the first information entity represents an information entity involved in the information associated with the query text;

[0273] A search unit, configured to filter out at least one second information entity associated with the first information entity from a pre-constructed knowledge graph; determine from the knowledge graph the information corresponding to each information entity and the association relationship between each information entity, wherein each information entity includes each first information entity and each second information entity;

[0274] The determining unit is further used to determine the display mode of each information entity, the information corresponding to each information entity, and the association relationship between each information entity according to a predetermined rule, and provide them to a display interface that needs to be displayed.

[0275] Optionally, the determination unit is specifically configured to: perform entity recognition on the first query text to obtain at least one third information entity; and determine a first information entity associated with the first query text according to each of the third information entities.

[0276] Optionally, the determination unit is specifically used to: perform entity expansion on the third information entity based on a pre-constructed knowledge graph to obtain at least one extended entity associated with the third information entity; determine a first information entity associated with the first query text according to the third information entity and the extended entity.

[0277] Optionally, the acquisition unit is further used to: acquire a second query text;

[0278] The determination unit is specifically used to: search for target information matching the second query text from the information stored in the pre-constructed knowledge graph; determine the information entities associated with the target information from the knowledge graph; and perform entity expansion on the third information entity based on the information entities associated with the target information; wherein the first query text is the text expanded on the basis of the second query text.

[0279] Optionally, the knowledge graph stores stored information and events, locations and people corresponding to the stored information;

[0280] The search unit is specifically used to: determine each target event involved in the target information; search for each target occurrence location and each target person associated with the target event from the knowledge graph; and determine the target event, the target occurrence location and the target person as information entities associated with the target information.

[0281] Optionally, the search unit is specifically used to: determine each target event involved in the target information from the knowledge graph; or, determine each target event involved in the target information from a pre-created event table, wherein the event table stores the correspondence between the information identifier of each stored information and the event.

[0282] Optionally, the device further comprises:

[0283] A grouping unit, used for grouping each of the third information entities and each of the extended entities to obtain a plurality of entity combinations, wherein the entity combinations include a plurality of information entities, and the information entities included in every two of the entity combinations are different;

[0284] An event reasoning unit, used for inputting the multiple groups of entity combinations into an event reasoning model to obtain events associated with the multiple groups of entity combinations;

[0285] The determining unit is further used to determine a display mode of events associated with the multiple groups of entity combinations, and provide the display mode to the display interface. The event reasoning model is trained based on sample entity combinations and sample events associated with the sample entity combinations.

[0286] Optionally, the determination unit is specifically used to determine the display method of each information entity, the information corresponding to each information entity, and the association relationship between the information entities according to the structure of the knowledge graph.

[0287] Optionally, the device further comprises:

[0288] The reference unit is used to determine the reference information corresponding to the first query text from a pre-constructed first reference information library, and provide the reference information associated with the query text to the display interface, wherein the reference information is used to provide knowledge reference for users to edit manuscripts.

[0289] Optionally, the device further comprises:

[0290] An updating unit is used to obtain update information and update the knowledge graph based on the update information.

[0291] Optionally, the knowledge graph includes multiple information nodes and multiple entity nodes, the information nodes store stored information, and the entity nodes store information entities corresponding to the stored information; the information entities corresponding to the stored information include events corresponding to the stored information; there is a first connecting edge between every two information nodes corresponding to the same event in the knowledge graph, and the first connecting edge stores events corresponding to the information nodes connected to the connecting edge.

[0292] Optionally, the multiple entity nodes include multiple event nodes, and the event nodes store events corresponding to the stored information. There is a second connecting edge between every two event nodes with an associated relationship, and the second connecting edge stores the association relationship between two events corresponding to the two connected event nodes.

[0293] Optionally, the multiple entity nodes further include multiple character nodes, the character nodes store character information corresponding to the stored information, a third connection edge exists between every two character nodes having an associated relationship, the third connection edge stores the association relationship between two characters corresponding to the two connected character nodes, a fourth connection edge exists between the character node and an event node or an information node associated with the character node, the fourth connection edge stores the association relationship between the connected character node and the information node or the event node;

[0294] And / or, the multiple entity nodes also include multiple location nodes, the location nodes store the occurrence location information corresponding to the stored information, there is a fifth connection edge between every two location nodes with an associated relationship, the fifth connection edge stores the association relationship between the two locations corresponding to the two connected location nodes, there is a sixth connection edge between the location node and the event node or information node associated with the location node, the sixth connection edge stores the association relationship between the connected location node and the information node or event node.

[0295] Optionally, the device further comprises:

[0296] A graph construction unit is used to obtain each information to be stored in the knowledge graph to be created; aggregate the information to be stored to obtain multiple groups of information groups, wherein each information in each information group is similar information; determine each information entity corresponding to the information group according to the content of each information in the information group; and create the knowledge graph based on each information entity corresponding to each information group and each information to be stored.

[0297] Optionally, the graph construction unit is specifically used to: aggregate each of the information to be stored based on an aggregation rule to obtain multiple groups of information groups, wherein the aggregation rule is: the content similarity between each information in each group of the information groups is greater than a first preset threshold, and the occurrence time interval is less than a preset time interval.

[0298] Corresponding to the search method based on the knowledge graph provided in the second embodiment of the present application, the fourth embodiment of the present application provides a search device based on the knowledge graph, which is applied to a client, and the device includes:

[0299] A second acquisition unit, acquiring a first query text input by a user, wherein the first query text is used for searching;

[0300] A display unit, used to obtain the display mode of each information entity, the information corresponding to each information entity, and the association relationship between the information entities, and display them on a display interface;

[0301] Among them, each information entity includes a first information entity and a second information entity, the first information entity is associated with the query text, the first information entity represents the information entity involved in the information associated with the query text, the second information entity is an information entity screened out from a pre-constructed knowledge graph and associated with the first information entity, and the information corresponding to each information entity and the association relationship between each information entity are respectively determined from the knowledge graph.

[0302] Optionally, the display unit is further used for:

[0303] Reference information associated with the query text is displayed on a display interface. The reference information is information associated with the first query text determined from a pre-constructed first reference information library. The reference information is used to provide knowledge reference for users to edit manuscripts.

[0304] Corresponding to the search method provided in the first embodiment of the present application, the fifth embodiment of the present application further provides an electronic device for searching information. Figure 7 As shown, the electronic device includes: a processor 701; and a memory 702, which is used to store a program of the search method. After the device is powered on and the program of the search method is run by the processor, the following steps are performed:

[0305] Get the first query text;

[0306] Determine at least one first information entity associated with the first query text, where the first information entity is used to represent an information entity involved in the information corresponding to the first query text;

[0307] Filtering at least one second information entity associated with the first information entity from a pre-constructed knowledge graph;

[0308] Determine from the knowledge graph the information corresponding to each information entity and the association relationship between each information entity, wherein each information entity includes each first information entity and each second information entity;

[0309] The display mode of each information entity, the information corresponding to each information entity, and the association relationship between each information entity is determined according to a predetermined rule, and provided to a display interface that needs to be displayed.

[0310] Corresponding to the search method provided in the second embodiment of the present application, the sixth embodiment of the present application further provides an electronic device for searching information. The electronic device includes: a processor; and a memory for storing a program of the search method. After the device is powered on and the program of the search method is run by the processor, the following steps are performed:

[0311] Acquire a first query text input by a user, where the first query text is used for information search;

[0312] Obtaining a display method of each information entity, information corresponding to each information entity, and association relationships between each information entity, and displaying them on a display interface;

[0313] Among them, each information entity includes a first information entity and a second information entity, the first information entity is associated with the query text, the first information entity represents the information entity involved in the information associated with the query text, the second information entity is an information entity screened out from a pre-constructed knowledge graph and associated with the first information entity, and the information corresponding to each information entity and the association relationship between each information entity are respectively determined from the knowledge graph.

[0314] Corresponding to the search method provided in the first embodiment of the present application, the seventh embodiment of the present application provides a computer-readable storage medium storing a program of the search method, the program is executed by a processor to perform the following steps:

[0315] Get the first query text;

[0316] Determine at least one first information entity associated with the first query text, where the first information entity is used to represent an information entity involved in the information corresponding to the first query text;

[0317] Filtering at least one second information entity associated with the first information entity from a pre-constructed knowledge graph;

[0318] Determine from the knowledge graph the information corresponding to each information entity and the association relationship between each information entity, wherein each information entity includes each first information entity and each second information entity;

[0319] The display mode of each information entity, the information corresponding to each information entity, and the association relationship between each information entity is determined according to a predetermined rule, and provided to a display interface that needs to be displayed.

[0320] Corresponding to the search method provided in the second embodiment of the present application, the eighth embodiment of the present application provides a computer-readable storage medium storing a program of the search method, the program is executed by a processor to perform the following steps:

[0321] Acquire a first query text input by a user, where the first query text is used for information search;

[0322] Obtaining a display method of each information entity, information corresponding to each information entity, and association relationships between each information entity, and displaying them on a display interface;

[0323] Among them, each information entity includes a first information entity and a second information entity, the first information entity is associated with the query text, the first information entity represents the information entity involved in the information associated with the query text, the second information entity is an information entity screened out from a pre-constructed knowledge graph and associated with the first information entity, and the information corresponding to each information entity and the association relationship between each information entity are respectively determined from the knowledge graph.

[0324] Corresponding to the search methods provided in the first and second embodiments of the present application, the ninth embodiment of the present application further provides a system for searching information, such as Figure 8 As shown, the system includes a server device 801 and a client device 802 .

[0325] The server device 801 is used to perform the following steps:

[0326] Obtaining a first query text from the client device 802;

[0327] Determine at least one first information entity associated with the first query text, where the first information entity is used to represent an information entity involved in the information corresponding to the first query text;

[0328] Filtering at least one second information entity associated with the first information entity from a pre-constructed knowledge graph;

[0329] Determine from the knowledge graph the information corresponding to each information entity and the association relationship between each information entity, wherein each information entity includes each first information entity and each second information entity;

[0330] The display method of each information entity, the information corresponding to each information entity, and the association relationship between each information entity is determined according to a predetermined rule, and provided to the client device 802.

[0331] The client device 802 is used to perform the following steps:

[0332] Acquire a first query text input by a user, and send the first query text to the server device 801, wherein the first query text is used for information search;

[0333] Receive each information entity sent by the server device 801, the information corresponding to each information entity, the association relationship between the information entities, and the display method of each received information, and display the information received from the server device 801 on the display interface according to the display method.

[0334] Among them, each information entity includes a first information entity and a second information entity, the first information entity is associated with the query text, the first information entity represents the information entity involved in the information associated with the query text, and the second information entity is an information entity screened from a pre-constructed knowledge graph and associated with the first information entity, and the information corresponding to each information entity and the association relationship between the information entities are respectively determined from the knowledge graph.

[0335] It should be noted that, for the methods, devices, electronic devices, and computer-readable storage medium embodiments provided in the second to eighth embodiments of the present application, detailed descriptions may refer to the relevant descriptions of the first embodiment of the present application, and will not be repeated here.

[0336] Although the present application is disclosed as above with preferred embodiments, it is not intended to limit the present application. Any technical personnel in this field may make possible changes and modifications without departing from the spirit and scope of the present application. Therefore, the scope of protection of the present application shall be based on the scope defined by the claims of the present application.

[0337] In a typical configuration, an electronic device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0338] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0339] 1. Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), random access memory (RAM) of other attributes, read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage media or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined in this article, computer-readable media does not include non-transitory media such as modulated data signals and carrier waves.

[0340] 2. Those skilled in the art should understand that the embodiments of the present application can be provided as methods, systems or computer program products. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0341] Although the present application is disclosed as above with preferred embodiments, it is not intended to limit the present application. Any technical personnel in this field may make possible changes and modifications without departing from the spirit and scope of the present application. Therefore, the scope of protection of the present application shall be based on the scope defined by the claims of the present application.

Claims

1. A search method based on knowledge graph, characterized in that: Applied to the server, the method includes: Get the first query text; Determine at least one first information entity associated with the first query text, the first information entity is used to represent an information entity involved in the information corresponding to the first query text, the first information entity is obtained according to the first query text and an extended entity, the extended entity is obtained according to each information entity associated with target information, the target information is information matching the second query text, and the first query text is text obtained by expanding the second query text; Filtering at least one second information entity associated with the first information entity from a pre-constructed knowledge graph; Determine from the knowledge graph the information corresponding to each information entity and the association relationship between each information entity, wherein each information entity includes each first information entity and each second information entity; The display mode of each information entity, the information corresponding to each information entity, and the association relationship between each information entity is determined according to a predetermined rule, and provided to a display interface that needs to be displayed.

2. The search method according to claim 1, characterized in that: The determining of the first information entity associated with the first query text comprises: Performing entity recognition on the query text first query text to obtain at least one third information entity; A first information entity associated with the first query text is determined according to each of the third information entities.

3. The search method according to claim 2, characterized in that: The determining, according to each of the third information entities, the first information entity associated with the first query text comprises: Performing entity expansion on the third information entity based on a pre-built knowledge graph to obtain at least one expanded entity associated with the third information entity; A first information entity associated with the first query text is determined according to the third information entity and the extended entity.

4. The search method according to claim 3, characterized in that: Before obtaining the first query text, the method further includes: Get the second query text; Searching for target information matching the second query text from each piece of information stored in the pre-built knowledge graph; Determine each information entity associated with the target information from the knowledge graph; The entity expansion of the third information entity based on the pre-built knowledge graph includes: Performing entity expansion on the third information entity based on each information entity associated with the target information; The first query text is a text obtained by expanding the second query text.

5. The search method according to claim 4, characterized in that: The knowledge graph stores stored information and events, locations and people corresponding to the stored information; The determining, from the knowledge graph, each information entity associated with the target information includes: Determining each target event involved in the target information; Searching for each target occurrence location and each target person associated with the target event from the knowledge graph; The target event, the target occurrence location, and the target person are determined as information entities associated with the target information.

6. The search method according to claim 3, characterized in that: The method further comprises: Grouping the third information entities and the extended entities to obtain multiple entity combinations, each entity combination containing multiple information entities, and each two entity combinations containing different information entities; Inputting the plurality of groups of entity combinations into an event reasoning model to obtain events associated with the plurality of groups of entity combinations; Determine the display mode of the events associated with the multiple groups of entity combinations and provide them to the display interface. The event reasoning model is trained based on sample entity combinations and sample events associated with the sample entity combinations.

7. The search method according to claim 1, characterized in that: The method of determining the display mode of each information entity, the information corresponding to each information entity, and the association relationship between each information entity according to a predetermined rule includes: The display method of each information entity, the information corresponding to each information entity, and the relationship between each information entity is determined according to the structure of the knowledge graph.

8. The search method according to claim 1, characterized in that: The method further comprises: Reference information corresponding to the first query text is determined from a pre-constructed first reference information library, and the reference information associated with the query text is provided to the display interface, where the reference information is used to provide knowledge reference for users to edit manuscripts.

9. The search method according to claim 1, characterized in that: The method further comprises: Obtain update information, and update the knowledge graph based on the update information.

10. The search method according to claim 1, characterized in that: The knowledge graph includes a plurality of information nodes and a plurality of entity nodes, the information nodes store stored information, and the entity nodes store information entities corresponding to the stored information; The information entity corresponding to the stored information includes the event corresponding to the stored information; There is a first connecting edge between every two information nodes corresponding to the same event in the knowledge graph, and the first connecting edge stores events corresponding to the information nodes connected to the connecting edge.

11. The search method according to claim 10, characterized in that: The multiple entity nodes include multiple event nodes, and the event nodes store events corresponding to the stored information. There is a second connecting edge between every two event nodes with an associated relationship, and the second connecting edge stores the association relationship between two events corresponding to the two connected event nodes.

12. The search method according to claim 1, characterized in that: The construction process of the knowledge graph includes: Obtain each piece of information to be stored in the knowledge graph to be created; Aggregating the information to be stored to obtain multiple information groups, wherein each information in each information group is similar information; Determine each information entity corresponding to the information group according to the content of each information in the information group; The knowledge graph is created based on the information entities corresponding to the information groups and the information to be stored.

13. The search method according to claim 12, characterized in that: The information to be stored is aggregated to obtain multiple groups of information groups, including: The information to be stored is aggregated based on an aggregation rule to obtain multiple information groups, wherein the aggregation rule is: the content similarity between each information in each information group is greater than a first preset threshold, and the occurrence time interval is less than a preset time interval.

14. A search method based on knowledge graph, characterized in that: Applied to a client, the method comprises: Acquire a first query text input by a user, where the first query text is used for information search; Obtaining a display method of each information entity, information corresponding to each information entity, and association relationships between each information entity, and displaying them on a display interface; Among them, each information entity includes a first information entity and a second information entity, the first information entity is associated with the query text, the first information entity represents the information entity involved in the information associated with the query text, the second information entity is an information entity screened from a pre-constructed knowledge graph and associated with the first information entity, the information corresponding to each information entity and the association relationship between the information entities are respectively determined from the knowledge graph, the first information entity is obtained according to the first query text and the extended entity, the extended entity is obtained according to each information entity associated with the target information, the target information is information matching the second query text, and the first query text is a text expanded on the basis of the second query text.

Images