CN113254756B

CN113254756B - Advertisement recall method, device, equipment and storage medium

Info

Publication number: CN113254756B
Application number: CN202010088191.7A
Authority: CN
Inventors: 王雪琼; 徐奋飞; 徐伟; 曹海军; 冯峤
Original assignee: Beijing Baidu Netcom Science and Technology Co Ltd
Current assignee: Beijing Baidu Netcom Science and Technology Co Ltd
Priority date: 2020-02-12
Filing date: 2020-02-12
Publication date: 2024-03-26
Anticipated expiration: 2040-02-12
Also published as: CN113254756A

Abstract

The embodiment of the application discloses an advertisement recall method, an advertisement recall device, advertisement recall equipment and a storage medium, and relates to the technical field of intelligent search. The specific implementation scheme is as follows: determining target entity nodes associated with the historical search request and the historical recall advertisement in a target domain knowledge graph and target attribute nodes connected with the target entity nodes; according to the target attribute node, determining the correlation between the historical search request and the historical recall advertisement; and determining a blacklist advertisement according to the correlation between the historical search request and the historical recall advertisement, and shielding advertisements with the correlation lower than a threshold value when the advertisement is recalled. The target field knowledge graph is applied to advertisement recall of a search scene to detect search requests without target field correlation and advertisements recalled by the search requests in historical search, so that effective guidance is carried out for subsequent advertisement recall, irrelevant advertisements are completely shielded, the accuracy of advertisement recall is improved, and the acquisition requirement of a user on target field information is met.

Description

Advertisement recall method, device, equipment and storage medium

Technical Field

The embodiment of the application relates to the technical field of computers, in particular to the technical field of intelligent search, and specifically relates to an advertisement recall method, an advertisement recall device, advertisement recall equipment and a storage medium.

Background

With the rapid development of the internet market, advertisements can be recalled for recommendation to users according to search requests initiated by the users, wherein advertisement recall in the medical field can be involved, namely, medical related advertisements are recalled according to medical search intention in the user search requests. At present, the conventional matching filtering mode cannot effectively shield all irrelevant advertisements, so that the accuracy of advertisement recall is low, and the user experience is affected.

Disclosure of Invention

The embodiment of the application provides an advertisement recall method, device, equipment and storage medium, which can improve the accuracy of advertisement recall.

In a first aspect, an embodiment of the present application provides an advertisement recall method, including:

determining a target entity node associated with a historical search request and a historical recall advertisement in a target domain knowledge graph, and a target attribute node connected with the target entity node;

determining the correlation between the historical search request and the historical recall advertisement according to the target attribute node;

and determining a blacklist advertisement according to the correlation between the historical search request and the historical recall advertisement, and shielding advertisements with the correlation lower than a threshold value when the advertisement is recalled.

One embodiment of the above application has the following advantages or benefits: the target field knowledge graph is applied to advertisement recall of a search scene to detect search requests without target field correlation and advertisements recalled by the search requests in historical search, so that effective guidance is carried out for subsequent advertisement recall, irrelevant advertisements are completely shielded, the accuracy of advertisement recall is improved, and the acquisition requirement of a user on target field information is met.

Optionally, the determining the target entity node associated with the historical search request and the historical recall advertisement in the target domain knowledge graph includes:

word segmentation processing is carried out on the historical search request and the target text in the historical recall advertisement;

according to the word segmentation result of the target text and the matching result between the nodes in the target domain knowledge graph, determining the core word and the candidate entity node of the target text;

and selecting a target entity node from the candidate entity nodes according to the core word.

One embodiment of the above application has the following advantages or benefits: the text of the historical search request and the historical recall advertisement is segmented and matched with the target domain knowledge graph to determine the target entity node corresponding to the text core word, so that the core entity content of the text can be obtained through the mapping of the target domain knowledge graph, and a basis is provided for the determination of the correlation between the historical search request and the historical recall advertisement.

Optionally, determining the core word of the target text according to the word segmentation result of the target text and the matching result between the nodes in the target domain knowledge graph includes:

and selecting core words of the target text from word segmentation results of the target text according to the confidence level of the word segmentation of the target domain knowledge graph in the target text.

One embodiment of the above application has the following advantages or benefits: based on the word segmentation result of the knowledge graph in the hit target field, the word segmentation method is favorable for selecting and obtaining the word segmentation which can best embody the text core content through the confidence degree of the word segmentation in the text environment.

Optionally, determining the candidate entity node of the target text according to the word segmentation result of the target text and the matching result between the nodes in the target domain knowledge graph includes:

matching the core word with the natural language word nodes in the target field knowledge graph; the natural language word nodes are trigger matching word representations of the nodes in the target field knowledge graph;

and according to the matching result of the natural language word nodes and the connection relation between the entity nodes and the attribute nodes in the target domain knowledge graph, taking the entity nodes associated with the core word as candidate entity nodes of the target text.

One embodiment of the above application has the following advantages or benefits: the matching triggering of the knowledge graph is carried out through the natural language word nodes in the knowledge graph of the target field, the natural language word nodes are used as the minimum matching granularity, the entity nodes of the root node are obtained by matching from bottom to top based on the connection relation between the nodes, and the maximum range of the entity nodes or the entity nodes can be included for representing various types of the attribute nodes in the natural language, so that the matching strength of the knowledge graph is improved, and the condition that the knowledge graph cannot be matched is avoided.

Optionally, after determining the blacklist advertisement according to the correlation between the historical search request and the historical recall advertisement, the method further comprises:

determining candidate recall advertisements according to the received current search request;

performing word segmentation processing on the current search request, and reconstructing according to a word segmentation result to obtain candidate search corpus;

and if the current search request or the candidate search corpus is detected to be the same as the historical search request, matching the candidate recall advertisement with the blacklist advertisement so as to shield the candidate recall advertisement hitting the blacklist advertisement.

One embodiment of the above application has the following advantages or benefits: through the determination of the blacklist advertisement, when the advertisement is recalled online, the current search request and the deformation form thereof are matched with the historical search request, the blacklist advertisement during periodic detection is shielded to the greatest extent, the advertisement which is irrelevant or has low relevance is prevented from being recommended to the user, and the advertisement recall accuracy and the user experience are improved.

Optionally, the target domain knowledge graph is a medical domain knowledge graph.

One embodiment of the above application has the following advantages or benefits: in the medical scene, the application of the medical knowledge graph in advertisement searching is beneficial to providing advertisements with strong correlation with medical contents searched by users for the users, and avoiding the wrong recall of the advertisements caused by the common medical attribute.

Optionally, before determining the target entity node associated with the historical search request and the historical recall advertisement in the target domain knowledge graph and the target attribute node connected with the target entity node, the method further includes:

according to the acquired disease information, taking the disease as an entity node and other disease information as attribute nodes, and constructing a medical knowledge graph taking the entity node as a core.

One embodiment of the above application has the following advantages or benefits: by constructing a medical knowledge graph with diseases as entity nodes and other disease information as attribute nodes, the method is beneficial to matching any disease information to obtain corresponding disease entities so as to detect relevance based on the disease entities.

Optionally, the constructing a medical knowledge graph with the entity node as a core includes:

Carrying out synonymous relation mining on the nodes in the medical knowledge graph;

and fusing different entity nodes according to at least one of the name attribute, the alias attribute and the synonymous relation of the entity nodes to obtain a redundancy-removed medical knowledge graph.

One embodiment of the above application has the following advantages or benefits: in the process of constructing the medical knowledge graph, the synonymous relation mining is carried out on the nodes, so that the method is beneficial to determining multiple word representations of the same node. Meanwhile, through fusion of different entity nodes in the medical knowledge graph and reservation of the fused entity nodes in other attribute forms, the redundancy of the medical knowledge graph is removed while the matching range of the knowledge graph is enlarged.

Optionally, the determining, according to the target attribute node, a correlation between the historical search request and the historical recall advertisement includes:

according to the core attribute node connected with the target entity node, determining the correlation between the historical search request and the historical recall advertisement; the core attribute node at least comprises a name attribute, a department attribute and a symptom attribute.

One embodiment of the above application has the following advantages or benefits: in the medical scene, the core attribute node is used as a node capable of absolutely distinguishing disease entities, so that the correlation between the historical search request and the historical recall advertisement is determined based on the core attribute node.

In a second aspect, an embodiment of the present application provides an advertisement recall apparatus, including:

the node matching module is used for determining target entity nodes associated with the historical search request and the historical recall advertisement in the target domain knowledge graph and target attribute nodes connected with the target entity nodes;

the advertisement relevance determining module is used for determining relevance between the historical search request and the historical recall advertisement according to the target attribute node;

and the advertisement shielding module is used for determining a blacklist advertisement according to the correlation between the historical search request and the historical recall advertisement and shielding advertisements with the correlation lower than a threshold value during advertisement recall.

In a third aspect, an embodiment of the present application provides an electronic device, including:

at least one processor; and

a memory communicatively coupled to the at least one processor; wherein,

The memory stores instructions executable by the at least one processor to enable the at least one processor to perform the advertising recall method described in any of the embodiments of the present application.

In a fourth aspect, embodiments of the present application provide a non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the advertising recall method of any embodiment of the present application.

One embodiment of the above application has the following advantages or benefits: based on the historical recall advertisements of the historical search request and the historical recall advertisements, determining entity nodes associated with the historical search request and the historical recall advertisements based on the target field knowledge graph, and determining the correlation between the historical search request and the historical recall advertisements according to the associated entity nodes, so that the historical recall advertisements with the correlation lower than a threshold value are added into a blacklist, and advertisements in the blacklist are shielded when the same search request is initiated again. According to the method and the device for detecting the advertisement recall of the target field, the target field knowledge graph is applied to the advertisement recall of the search scene, so that search requests which do not have target field correlation in historical search and advertisements recalled by the search requests are detected, effective guidance is conducted on subsequent advertisement recalls, irrelevant advertisements are completely shielded, accuracy of the advertisement recall is improved, and the requirement of a user for acquisition of target field information is met.

Other effects of the above alternative will be described below in connection with specific embodiments.

Drawings

The drawings are for better understanding of the present solution and do not constitute a limitation of the present application. Wherein:

FIG. 1 is a flow chart of an advertising recall method according to a first embodiment of the present application;

FIG. 2 is a flow chart of an advertising recall method according to a second embodiment of the present application;

FIG. 3 is a partial exemplary diagram of a medical knowledge-graph according to a second embodiment of the present application;

FIG. 4 is a block diagram of an advertisement recall based on medical knowledge-graph according to a second embodiment of the present application;

FIG. 5 is a flow chart of an advertising recall method according to a third embodiment of the present application;

FIG. 6 is a flow chart of an advertising recall method according to a fourth embodiment of the present application;

FIG. 7 is a schematic diagram of an advertising recall device according to a fifth embodiment of the present application;

FIG. 8 is a block diagram of an electronic device used to implement the advertisement recall method of an embodiment of the present application.

Detailed Description

Exemplary embodiments of the present application are described below in conjunction with the accompanying drawings, which include various details of the embodiments of the present application to facilitate understanding, and should be considered as merely exemplary. Accordingly, one of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the present application. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.

First embodiment

Fig. 1 is a flowchart of an advertisement recall method according to a first embodiment of the present application, which is applicable to periodically detecting the correlation between a historical search request initiated by a user and a historical recall advertisement recalled by the user based on a target domain knowledge graph, and determining a blacklist advertisement with a correlation lower than a threshold value, so as to shield an irrelevant advertisement from an online advertisement recall. As shown in fig. 1, the method specifically includes the following steps:

s110, determining target entity nodes associated with the historical search request and the historical recall advertisement in the target domain knowledge graph and target attribute nodes connected with the target entity nodes.

In particular embodiments of the present application, the target area may be any area relevant to the search, such as a medical area, and the like. Historical search requests refer to search requests initiated in a target area by a large number of users in a network over a period of time. Accordingly, the historical recall advertisement refers to an advertisement corresponding to recall by the advertisement platform based on the historical search request. Wherein, the correlation between different historical recall advertisements and the same historical search request is different, the stronger the correlation is, the more the historical recall advertisement meets the historical search requirement of the user.

In this embodiment, the target domain knowledge graph is constructed to be applied to advertisement recall according to a search request, aiming at the technical problem that in the prior art, when only advertisements are recalled based on keyword exact matching, effective shielding cannot be performed on all irrelevant advertisements. The target domain knowledge graph is composed of at least entity nodes and attribute nodes, and uses the entity nodes as cores and the attribute nodes as concrete explanation of the entity nodes, so that the target domain knowledge graph can be matched with the associated entity nodes according to the attribute nodes. Wherein different entity nodes may have the same attribute node.

In addition, the target domain knowledge graph can also comprise natural language word nodes, the natural language word nodes are in one-to-one correspondence with entity nodes or attribute nodes in the target domain knowledge graph, are word representations in different forms of the corresponding entity nodes or attribute nodes, can contain multiple representations of the entity nodes or attribute nodes in natural language in a maximum range, can serve as the minimum matching granularity of the target domain knowledge graph, and are used for triggering node matching so as to improve the matching strength of the knowledge graph and avoid the situation of incapability of matching.

Illustratively, in the medical field, nodes in the medical knowledge graph are determined according to disease information collected from the website of the authority medical institution, with the disease as an entity node and other disease information as attribute nodes. Respectively carrying out synonymous relation mining on entity nodes and attribute nodes in the medical knowledge graph; and fusing different entity nodes according to at least one of the name attribute, the alias attribute and the synonymous relation of the entity nodes to obtain a medical knowledge graph which not only contains the dimension information but also is redundant. Wherein, the natural language word nodes serving as the corresponding nodes can be also represented by folk words with various diseases or attributes. Therefore, the matching can be triggered based on the natural language word nodes, the attribute nodes are determined according to the matched natural language word nodes, and the entity nodes connected with the attribute nodes are determined according to the connection relation between the nodes in the target domain knowledge graph.

Specifically, at least one of a title, a text or a picture text in a historical search request and a historical recall advertisement can be used as a target text to be matched, word segmentation processing is performed on the target text, each word obtained is matched with a node in a target domain knowledge graph, and a core word and a candidate entity node which hit the target domain knowledge graph in the target text and can embody the core content of the target text are determined. The core word of the target text can be selected from word segmentation results of the target text according to the confidence level in the target text of the word segmentation of the target domain knowledge graph. And matching the core word with the natural language word nodes in the target domain knowledge graph, and taking the entity nodes associated with the core word as candidate entity nodes of the target text. Finally, the relevance between the core word and the candidate entity node can be calculated based on the relevance model, the candidate entity node with lower relevance is filtered, and the candidate entity node with highest relevance and meeting a certain threshold is used as a final target entity node.

S120, according to the target attribute node, determining the correlation between the historical search request and the historical recall advertisement.

In the specific embodiment of the application, after the target entity node of the history search request and the associated target attribute node thereof and the target entity node of the history recall advertisement and the associated target attribute node thereof are determined, the correlation between the history search request and the history recall advertisement can be determined based on the target attribute nodes of the history search request and the history recall advertisement.

For example, attribute nodes of entity nodes may be divided into core attribute nodes and non-core attribute nodes based on characteristics of entities in the target domain in advance. The core attribute node is generally a mark attribute of the entity node, and has better differentiation to the entity; non-core attribute nodes are attributes for which distinguishing is not so obvious. And determining the correlation between the historical search request and the historical recall advertisement based on the core attribute node connected with the target entity node.

Illustratively, in the medical field, a map structure for example of cold disease is shown in table 1. The core attribute node at least comprises a name attribute, a department attribute and a symptom attribute. Assuming that the target entity node of the history search request is determined to be A, and the target entity node of a certain history recall advertisement is determined to be B. Based on the medical knowledge graph, if it is detected that the department attribute nodes of a and B are the same or have a superior-inferior relationship, or if it is detected that a and B are the same disease entity node or that the department attribute nodes of a and B have an intersection on the department attribute, or if it is detected that the department of one disease is contained in the department information of another disease, it may be determined that the history search request has a correlation with the history recall advertisement.

Table 1 exemplary table of map structures

Project	Value of	Sample example
			Primary category (category)	First-stage department	Internal medicine
Second category (category)	Second-level department	Respiratory medicine
			Entity (disease)	Signature of disease names
Core attribute (core_attr)	Disease name	Cold treating medicine
			Core attribute (core_attr)	Symptoms of	Sneeze
Core attribute (core_attr)	Department of science	Respiratory medicine
			Uncore coreAttribute (nocore_attr)	Part(s)	Nose
Non-core properties (nocore_attr)	Administration of drugs	Radix Isatidis
			Non-core properties (nocore_attr)	Clinical examination	Nasal cavity stenosis
Non-core properties (nocore_attr)	Susceptible crowd	Children' s
			Non-core properties (nocore_attr)	Infection mode	Droplet propagation
Non-core properties (nocore_attr)	Therapeutic method	Drug treatment
			Natural language word node (motion)	Values and synonyms for attributes

S130, determining blacklist advertisements according to the correlation between the historical search request and the historical recall advertisements, and shielding advertisements with correlation lower than a threshold value during advertisement recall.

In a specific embodiment of the present application, the blacklist advertisement refers to a historical recall advertisement having no correlation with a historical search request or having a correlation lower than a threshold value, and is used for indicating that in a search system, when searching and advertisement recall are initiated again according to the historical search request or the recombination of word segmentation results of the historical search request, if the recall results include an advertisement in the blacklist, the advertisement is shielded online so as to avoid recommending the advertisement having the lower correlation to a user.

Specifically, by periodically performing the entity linking in S110 and the relevance discrimination in S120, the history recall advertisement whose relevance is lower than the threshold value, which is determined based on the target-area knowledge-graph detection, is added to the blacklist. In the blacklist, the history search request and the recombination of word segmentation results of the history search request can be established, and the association relationship between the history search request and the blacklist advertisement can be established.

According to the technical scheme of the embodiment, based on the historical search request and the historical recall advertisement recalled by the historical search request, the entity nodes associated with the historical search request and the historical recall advertisement are determined based on the target field knowledge graph, and the correlation between the historical search request and the historical recall advertisement is determined according to the associated entity nodes, so that the historical recall advertisement with the correlation lower than the threshold value is added into the blacklist, and the advertisement in the blacklist is shielded when the same search request is initiated again. According to the method and the device for detecting the advertisement recall of the target field, the target field knowledge graph is applied to the advertisement recall of the search scene, so that search requests which do not have target field correlation in historical search and advertisements recalled by the search requests are detected, effective guidance is conducted on subsequent advertisement recalls, irrelevant advertisements are completely shielded, accuracy of the advertisement recall is improved, and the requirement of a user for acquisition of target field information is met.

Second embodiment

Fig. 2 is a flowchart of an advertisement recall method according to a second embodiment of the present application, where, based on the first embodiment, determining, based on a target domain knowledge graph, a history search request and entity nodes associated with a history recall advertisement are further explained, and the entity nodes can be obtained by matching text core words determined based on a word segmentation result of text. As shown in fig. 2, the method specifically includes the following steps:

s210, word segmentation processing is carried out on the target text in the historical search request and the historical recall advertisement.

In the specific embodiment of the application, the historical search request text can be used as the target text, and word segmentation processing can be performed on the historical search request text. The advertisement title of the history recall advertisement or text content in the history recall advertisement, or even text content identified from advertisement pictures, can be used as a target text to perform word segmentation processing on the history recall advertisement text. The word segmentation processing method in this embodiment is not limited, and any method capable of performing word segmentation processing on text may be applied to this embodiment.

For example, assuming that the history search request is "what medicine is consumed by cold and cough", the text of the history search request may be used as a target text, and word segmentation processing may be performed on the target text to obtain word segmentation results of "cold", "cough", "eat", "what" and "medicine".

S220, determining core words and candidate entity nodes of the target text according to a word segmentation result of the target text and a matching result between nodes in the target field knowledge graph.

In the specific embodiment of the application, after word segmentation processing is performed on the target text, each word segmentation in the word segmentation result of the target text is respectively matched with a node in the target domain knowledge graph, and subsequent entity linking and correlation discrimination are performed on the basis of the matching result between the word segmentation result and the node in the target domain knowledge graph. In order to improve the matching capability of the knowledge graph in the target field, natural language word nodes are used as the minimum matching granularity for triggering the matching of the nodes.

Illustratively, taking a medical knowledge graph as an example, fig. 3 is a partial illustration of the medical knowledge graph. As shown in fig. 3, the disease is taken as an entity node, name, cough, symptom, medication, clinical examination, treatment method, susceptible crowd and transmission mode are taken as attribute nodes, and the motion label is taken as a natural language word node. For example, when matching based on disease names, whether the word in the target text is a multiple word representation of cold, catarrhal rhinitis, etc., the attribute node cold, and the disease entity node to which the attribute node is connected, can be matched based on the natural language word node.

In this embodiment, the core word refers to a word segment that best represents the core content of the target text. Optionally, according to the word segmentation hitting the target domain knowledge graph, selecting a core word of the target text from the word segmentation result of the target text according to the confidence level in the target text. The method of determining the confidence coefficient is not limited in this embodiment, and any method capable of determining the confidence coefficient may be applied to this embodiment.

The word segmentation method includes the steps that according to a word segmentation result of a target text and a matching result between nodes in a target domain knowledge graph, confidence levels of the words in the target text, which are the words in the target domain knowledge graph, are calculated based on a word rank technology, the words are ranked according to the confidence levels, and the word with the highest confidence level is used as a core word of the target text. For example, in the above example, in the word segmentation result, the words "cold", "cough", and "medicine" hit the medical knowledge map. And calculating the confidence of the word "cold", "cough" and "medicine" in the target text based on the wordrenk technology, so as to obtain the highest confidence of the word "cold" in the context of the historical search request, and using the word "cold" as the core word of the historical search request.

In this embodiment, the candidate entity nodes are entity nodes that can relatively represent the core content of the target text, among all entity nodes associated with the word segmentation of the target domain knowledge graph hit by the target text, and entity nodes with core words matched in the target domain knowledge graph can be used as candidate entity nodes of the target text. Optionally, matching the core word with the natural language word nodes in the knowledge graph of the target field; and according to the matching result of the natural language word nodes and the connection relation between the entity nodes and the attribute nodes in the target domain knowledge graph, taking the entity nodes associated with the core word as candidate entity nodes of the target text. According to the matching of the target domain knowledge graph, more than one entity node can be associated with the core word. For example, individual symptoms of different diseases may be identical, so that when matching is performed using the symptoms as core words, different disease entity nodes may be obtained by matching.

In this embodiment, in order to improve entity linking efficiency and avoid invalidating the history recall advertisement that does not relate to the target domain knowledge graph, after determining the word segmentation result of the target text of the history recall advertisement, when matching is performed based on the natural language word nodes, the history recall advertisement to which the target text of the target domain knowledge graph is not hit by all the words can be filtered.

S230, selecting a target entity node from the candidate entity nodes according to the core word.

In the specific embodiment of the application, the target entity node refers to an entity node which can best embody the core content of the target text. The relevance between the core word and the candidate entity node can be calculated based on the relevance model, the candidate entity node with lower relevance is filtered, and the candidate entity node with highest relevance and meeting a certain threshold is used as the final target entity node.

Illustratively, in the above example, in the word hitting the medical knowledge graph, it is assumed that the entity node associated with the word "cold" includes A, B and C, the entity node associated with the word "cough" includes D and E, and the entity node associated with the word "medicine" includes F. According to the confidence of each hit word in the target text, the word "cold" is determined as a core word, and the entity nodes A, B and C are taken as candidate entity nodes. And calculating the relevance between the core word and the candidate entity node based on the relevance model, and taking the candidate entity node with the highest relevance in A, B and C and meeting a certain threshold as the final target entity node.

S240, according to the target attribute node connected with the target entity node, determining the correlation between the historical search request and the historical recall advertisement.

S250, determining blacklist advertisements according to the correlation between the historical search request and the historical recall advertisements, and shielding advertisements with correlation lower than a threshold value during advertisement recall.

Illustratively, taking the medical field as an example, fig. 4 is a block diagram of a structure for advertisement recall based on a medical knowledge graph. As shown in FIG. 4, the authoritative medical website, the phoenix nest medical material and the disease information of the DMP (Data Management Platform ) are taken as disease data sources. In the medical knowledge graph platform, firstly, a medical knowledge graph is constructed based on a disease data source through entity construction, synonymous relation mining, entity fusion and natural language word node construction, so as to form a disease knowledge Base (KG-Base). When the historical recall advertisements are periodically detected, the medical knowledge graph platform performs entity linking on the target text through text matching filtering, entity identification, word rank and relevance filtering. And finally, matching the disease entity nodes linked with the historical search request and the historical recall advertisement, and determining the correlation between the historical search request and the historical recall advertisement so as to mine the blacklist advertisement, and shielding the recalled blacklist advertisement when the advertisement recalls.

According to the technical scheme, through word segmentation processing is conducted on the historical search request and the target text of the recalled historical recalled advertisement, node matching is conducted on the basis of the target domain knowledge graph, core words and candidate entity nodes of the historical search request and the historical recalled advertisement are determined, the target entity nodes are selected from the candidate entity nodes according to the core words, the correlation between the historical search request and the historical recalled advertisement is determined according to the target attribute nodes connected with the target entity nodes, and therefore the historical recalled advertisement with the correlation lower than the threshold value is added into a blacklist, and advertisements in the blacklist are shielded when the same search request is initiated again. According to the method and the device for detecting the advertisement recall of the target field, the target field knowledge graph is applied to the advertisement recall of the search scene, so that search requests which do not have target field correlation in historical search and advertisements recalled by the search requests are detected, effective guidance is conducted on subsequent advertisement recalls, irrelevant advertisements are completely shielded, accuracy of the advertisement recall is improved, and the requirement of a user for acquisition of target field information is met.

Third embodiment

Fig. 5 is a flowchart of an advertisement recall method according to a third embodiment of the present application, where, based on the first embodiment, the description is further given of correlation detection between a history search request and a history recall advertisement in a medical field, and a medical knowledge graph can be constructed based on disease information, and correlation detection can be performed based on the medical knowledge graph. As shown in fig. 5, the method specifically includes the following steps:

S510, constructing a medical knowledge graph with the solid node as a core by taking the diseases as the solid node and other disease information as attribute nodes according to the acquired disease information.

In the specific embodiment of the present application, in the medical field, the disease information is disease-related information collected or grasped in advance from an authoritative medical institution website, such as cases, papers, and the like. The nodes in the medical knowledge graph can be determined by identifying words such as diseases and attributes in the disease information in a keyword identification mode and the like. And establishing edges among the nodes according to medical field expertise, and determining the association relation among the nodes.

Optionally, carrying out synonymous relation mining on the nodes in the medical knowledge graph; and fusing different entity nodes according to at least one of the name attribute, the alias attribute and the synonymous relation of the entity nodes to obtain a redundancy-removed medical knowledge graph.

In this embodiment, the representation of the disease information is relatively specialized in view of the relative authority of the source of the disease information, and the specialized representation of the same word may include more than one type. Therefore, it is necessary to mine nodes with synonymous relationships in the medical knowledge graph. For the attribute nodes, the attribute nodes in the map can be subjected to synonym relation mining of attribute names according to a synonym table of a preset synonym relation attribute word so as to determine the attribute nodes with the synonym relation, and various professional expression forms of the attribute nodes are expanded. For entity nodes, the entity in the map can be subjected to synonymous relation mining according to the text similarity of the disease names and/or the number of the common attributes of the two entity association. For example, rhinitis and chronic rhinitis, while varying in the profession, have little to no significant variation for the user.

In this embodiment, the mining of the synonymous relationship realizes the full expansion of the expertise of the medical knowledge graph, and in order to improve the simplicity and the application efficiency of the medical knowledge graph, the entity nodes are required to be fused. Specifically, different entities and nodes meeting the following arbitrary relationship are fused: the entity names are the same; the alias attributes of the entity nodes have the same alias; two entities with synonymous relationships. When the entity nodes are fused, the entity nodes with more attribute labels are used as the fused entity nodes, the union of the associated attributes of the fused entity nodes is reserved, and the fused entity nodes are reserved in the form of alias attributes.

In addition, after medical knowledge graph construction is performed based on medical professional knowledge, natural language expression forms of all nodes can be extracted according to folk expressions to serve as natural language word nodes of all nodes. The method further reduces the professional requirement of the user for inputting the medical search request, the user can initiate search in any known word representation, and the matching of the medical knowledge graph can be performed according to any word representation.

S520, word segmentation processing is carried out on the target text in the historical search request and the historical recall advertisement.

S530, selecting core words of the target text from word segmentation results of the target text according to the confidence level in the target text of the word segmentation of the hit medical knowledge graph.

S540, matching the core word with the natural language word nodes in the medical knowledge graph; the natural language word nodes are trigger matching word representations of the nodes in the medical knowledge graph.

S550, according to the matching result of the natural language word nodes and the connection relation between the entity nodes and the attribute nodes in the medical knowledge graph, taking the entity nodes associated with the core word as candidate entity nodes of the target text.

S560, selecting a target entity node from the candidate entity nodes according to the core word.

S570, determining the correlation between the historical search request and the historical recall advertisement according to the core attribute node connected with the target entity node; the core attribute node at least comprises a name attribute, a department attribute and a symptom attribute.

S580, determining a blacklist advertisement according to the correlation between the historical search request and the historical recall advertisement, and shielding advertisements with the correlation lower than a threshold value when the advertisement is recalled.

According to the technical scheme, the medical knowledge graph taking the disease as the entity node and other disease information as the attribute node is constructed, and the medical knowledge graph is applied to advertisement recall of a search scene, so that search requests without medical relevance and advertisements recalled by the search requests are detected in a history search log, the subsequent advertisement recall is effectively guided, the uncorrelated advertisements are completely shielded, the accuracy of the advertisement recall is improved, and the acquisition requirement of a user on the medical information is met.

Fourth embodiment

Fig. 6 is a flowchart of an advertisement recall method according to a fourth embodiment of the present application, where the embodiment is further explained based on the online advertisement recall process of the blacklist advertisement based on the first embodiment, and can mask advertisements with a relevance lower than a threshold based on the blacklist advertisement. As shown in fig. 6, the method specifically includes the following steps:

s610, determining target entity nodes associated with the historical search request and the historical recall advertisement in a target domain knowledge graph and target attribute nodes connected with the target entity nodes.

S620, according to the target attribute node, determining the correlation between the historical search request and the historical recall advertisement.

S630, determining blacklist advertisements according to the correlation between the historical search request and the historical recall advertisements, and shielding advertisements with correlation lower than a threshold value during advertisement recall.

In the specific embodiment of the application, the correlation judgment can be periodically carried out on the historical search request and the historical recall advertisement recalled by the historical search request so as to determine the blacklist advertisement. Illustratively, at night zero every day, entity linking and relevance discrimination is performed with all historical search requests of all users of the same day and advertisement titles of recalled historical recall advertisements as target texts. And adding the historical search request without relevance or with reduced relevance and the associated historical recall advertisement thereof to the blacklist. For example, a history search request with a common cold as a subject and a history recall advertisement with an integer as a subject recalled by the history search request are added to a blacklist. And when any user initiates the search for the cold, if the recalled advertisement comprises the advertisement integer, namely the blacklist is hit, the recalled integer advertisement is filtered.

S640, determining candidate recall advertisements according to the received current search request.

In a specific embodiment of the present application, the current search request refers to a search request initiated by a user and received online in real time. Accordingly, all advertisements recalled according to the current search request are taken as candidate recall advertisements to be recommended to the user based on any advertisement recall mode in the prior art.

S650, word segmentation processing is carried out on the current search request, and reconstruction is carried out according to word segmentation results, so that candidate search corpus is obtained.

In the embodiment of the application, the text with the same semantic meaning is expressed in various ways and different users have different expression habits, so that word segmentation processing is performed on the current search request, and reconstruction is performed based on word segmentation results to obtain various expression forms for expressing the current search request, so that the candidate search corpus is formed.

And S660, if the current search request or the candidate search corpus is detected and is the same as the historical search request, matching the candidate recall advertisement with the blacklist advertisement so as to shield the candidate recall advertisement hitting the blacklist advertisement.

In the specific embodiment of the application, matching the current search request and the candidate search corpus with the historical search request, and detecting whether the same or similar search request is initiated historically. If the current search request or the candidate search corpus is detected to be the same as the historical search request, matching the candidate recall advertisement with the blacklist advertisement, and shielding the candidate recall advertisement hitting the blacklist advertisement under the guidance of the blacklist advertisement. And further overcomes the defect that the prior art cannot effectively shield all irrelevant advertisements.

According to the technical scheme, through the determination of the blacklist advertisement, when the advertisement is recalled online, the current search request and the deformation form thereof are matched with the historical search request, the blacklist advertisement during periodic detection is shielded to the greatest extent, the advertisement which is irrelevant or has low relevance is prevented from being recommended to the user, and the advertisement recall accuracy and the user experience are improved.

Fifth embodiment

Fig. 7 is a schematic structural diagram of an advertisement recall device according to a fifth embodiment of the present application, where the embodiment may be adapted to periodically detect, based on a target domain knowledge graph, a correlation between a historical search request initiated by a user and a historical recall advertisement recalled by the user, and determine a blacklist advertisement with a correlation lower than a threshold, so as to mask an irrelevant advertisement from an online advertisement recall. The apparatus 700 specifically includes the following:

the node matching module 710 is configured to determine a target entity node associated with the historical search request and the historical recall advertisement in the target domain knowledge graph, and a target attribute node connected with the target entity node;

an advertisement relevance determining module 720, configured to determine relevance between the historical search request and the historical recall advertisement according to the target attribute node;

And the advertisement shielding module 730 is configured to determine a blacklist advertisement according to the correlation between the historical search request and the historical recall advertisement, and to shield advertisements with correlation lower than a threshold value during advertisement recall.

Optionally, the node matching module 710 is specifically configured to:

Further, the apparatus 700 further includes an advertisement online recall module 740, specifically configured to:

after determining blacklist advertisements according to the correlation between the historical search requests and the historical recall advertisements, determining candidate recall advertisements according to the received current search requests;

Further, the apparatus 700 further includes a map construction module 750, specifically configured to:

and before determining a target entity node associated with the historical search request and the historical recall advertisement in the target domain knowledge graph and a target attribute node connected with the target entity node, constructing a medical knowledge graph with the entity node as a core by taking the disease as the entity node and other disease information as the attribute node according to the acquired disease information.

Optionally, the map construction module 750 is specifically configured to:

Optionally, the advertisement relevance determining module 720 is specifically configured to:

According to the technical scheme, through the mutual coordination among the functional modules, the functions of building a target field knowledge graph, linking entities, judging the correlation between a historical search request and a historical recall advertisement, determining a blacklist, carrying out real-time online recall of the advertisement, shielding the blacklist advertisement and the like are realized. According to the method and the device for detecting the advertisement recall of the target field, the target field knowledge graph is applied to the advertisement recall of the search scene, so that search requests which do not have target field correlation in historical search and advertisements recalled by the search requests are detected, effective guidance is conducted on subsequent advertisement recalls, irrelevant advertisements are completely shielded, accuracy of the advertisement recall is improved, and the requirement of a user for acquisition of target field information is met.

Sixth embodiment

According to embodiments of the present application, an electronic device and a readable storage medium are also provided.

As shown in fig. 8, a block diagram of an electronic device of an advertising recall method according to an embodiment of the present application. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular telephones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the application described and/or claimed herein.

As shown in fig. 8, the electronic device includes: one or more processors 801, memory 802, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The various components are interconnected using different buses and may be mounted on a common motherboard or in other manners as desired. The processor may process instructions executing within the electronic device, including instructions stored in or on memory to display graphical information of a graphical user interface (Graphical User Interface, GUI) on an external input/output device, such as a display device coupled to the interface. In other embodiments, multiple processors and/or multiple buses may be used, if desired, along with multiple memories and multiple memories. Also, multiple electronic devices may be connected, each providing a portion of the necessary operations, e.g., as a server array, a set of blade servers, or a multiprocessor system. One processor 801 is illustrated in fig. 8.

Memory 802 is a non-transitory computer-readable storage medium provided herein. Wherein the memory stores instructions executable by the at least one processor to cause the at least one processor to perform the advertisement recall method provided herein. The non-transitory computer readable storage medium of the present application stores computer instructions for causing a computer to perform the advertisement recall method provided by the present application.

The memory 802 is used as a non-transitory computer readable storage medium for storing non-transitory software programs, non-transitory computer executable programs, and modules, such as program instructions/modules corresponding to the advertisement recall method in the embodiments of the present application, for example, the node matching module 710, the advertisement relevance determining module 720, the advertisement mask module 730, the advertisement online recall module 740, and the map building module 750 shown in fig. 7. The processor 801 executes various functional applications of the server and data processing, i.e., implements the advertisement recall method in the above-described method embodiments, by running non-transitory software programs, instructions, and modules stored in the memory 802.

Memory 802 may include a storage program area that may store an operating system, at least one application program required for functionality, and a storage data area; the storage data area may store data created according to the use of the electronic device of the advertisement recall method, and the like. In addition, memory 802 may include high-speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory 802 may optionally include memory remotely located with respect to processor 801, which may be connected to the electronic device of the advertising recall method via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.

The electronic device of the advertisement recall method may further include: an input device 803 and an output device 804. The processor 801, memory 802, input devices 803, and output devices 804 may be connected by a bus or other means, for example in fig. 8.

The input device 803 may receive entered numeric or character information and generate key signal inputs related to user settings and function control of the electronic device of the advertising recall method, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, a pointer stick, one or more mouse buttons, a track ball, a joystick, and the like. The output means 804 may include a display device, auxiliary lighting means, such as light emitting diodes (Light Emitting Diode, LEDs), tactile feedback means, and the like; haptic feedback devices such as vibration motors and the like. The display device may include, but is not limited to, a liquid crystal display (Liquid Crystal Display, LCD), an LED display, and a plasma display. In some implementations, the display device may be a touch screen.

Various implementations of the systems and techniques described here can be implemented in digital electronic circuitry, integrated circuitry, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs, the one or more computer programs may be executed and/or interpreted on a programmable system including at least one programmable processor, which may be a special purpose or general-purpose programmable processor, that may receive data and instructions from, and transmit data and instructions to, a storage system, at least one input device, and at least one output device.

These computing programs, also referred to as programs, software applications, or code, include machine instructions for a programmable processor, and may be implemented in a high-level procedural and/or object-oriented programming language, and/or in assembly/machine language. As used herein, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, apparatus, and/or device for providing machine instructions and/or data to a programmable processor, e.g., magnetic discs, optical disks, memory, programmable logic devices (Programmable Logic Device, PLD), including a machine-readable medium that receives machine instructions as a machine-readable signal. The term "machine-readable signal" refers to any signal used to provide machine instructions and/or data to a programmable processor.

To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: a display device for displaying information to a user, for example, a Cathode Ray Tube (CRT) or an LCD monitor; and a keyboard and pointing device, such as a mouse or trackball, by which a user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user may be any form of sensory feedback, e.g., visual feedback, auditory feedback, or tactile feedback; and input from the user may be received in any form, including acoustic input, speech input, or tactile input.

The systems and techniques described here can be implemented in a computing system that includes a background component, e.g., a data server, or that includes a middleware component, e.g., an application server, or that includes a front-end component, e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here, or any combination of such background, middleware, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication, e.g., a communication network. Examples of communication networks include: local area network (Local Area Network, LAN), wide area network (Wide Area Network, WAN), the internet and blockchain networks.

The computer system may include a client and a server. The client and server are typically remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other.

According to the technical scheme of the embodiment of the application, the target field knowledge graph is applied to the advertisement recall of the search scene to detect the search request without target field correlation and the recalled advertisement thereof in the history search, so that effective guidance is carried out for the subsequent advertisement recall, the uncorrelated advertisement is completely shielded, the accuracy of the advertisement recall is improved, and the acquisition requirement of a user on the target field information is met.

In addition, through word segmentation of the text of the historical search request and the historical recall advertisement and matching with the target domain knowledge graph, target entity nodes corresponding to the text core words are determined, core entity contents of the text are obtained through mapping of the target domain knowledge graph, and a basis is provided for determining correlation between the historical search request and the historical recall advertisement.

In addition, based on the word segmentation result of the knowledge graph in the hit target field, the word segmentation method is favorable for selecting and obtaining the word segmentation which can best embody the text core content through the confidence degree of the word segmentation in the text environment.

In addition, the matching triggering of the knowledge graph is carried out through the natural language word nodes in the knowledge graph in the target field, the natural language word nodes are used as the minimum matching granularity, the entity nodes of the root node are obtained by matching from bottom to top based on the connection relation among the nodes, and the maximum range of the entity nodes or the entity nodes can be included for representing various types of attribute nodes in the natural language, so that the matching strength of the knowledge graph is improved, and the condition that the knowledge graph cannot be matched is avoided.

In addition, through the determination of the blacklist advertisement, when the advertisement is recalled online, the current search request and the deformation form thereof are matched with the historical search request, the blacklist advertisement during periodic detection is shielded to the greatest extent, the advertisement which is irrelevant or has low relevance is prevented from being recommended to the user, and the advertisement recall accuracy and the user experience are improved.

In addition, in the medical scene, the application of the medical knowledge graph in advertisement searching is beneficial to providing advertisements with strong correlation with medical content searched by the user for the user, and avoiding the wrong recall of the advertisements caused by the common medical attribute.

In addition, by constructing a medical knowledge graph with diseases as entity nodes and other disease information as attribute nodes, the method is beneficial to matching any disease information to obtain corresponding disease entities so as to detect relevance based on the disease entities.

In addition, in the process of constructing the medical knowledge graph, the synonymous relation mining is carried out on the nodes, so that the determination of multiple word representations of the same node is facilitated. Meanwhile, through fusion of different entity nodes in the medical knowledge graph and reservation of the fused entity nodes in other attribute forms, the redundancy of the medical knowledge graph is removed while the matching range of the knowledge graph is enlarged.

In addition, in the medical scene, the core attribute node is used as a node capable of absolutely distinguishing disease entities, so that the correlation between the historical search request and the historical recall advertisement is determined based on the core attribute node.

It should be appreciated that various forms of the flows shown above may be used to reorder, add, or delete steps. For example, the steps described in the present application may be performed in parallel, sequentially, or in a different order, provided that the desired results of the technical solutions disclosed in the present application can be achieved, and are not limited herein.

The above embodiments do not limit the scope of the application. It will be apparent to those skilled in the art that various modifications, combinations, sub-combinations and alternatives are possible, depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application are intended to be included within the scope of the present application.

Claims

1. An advertising recall method, comprising:

determining a blacklist advertisement according to the correlation between the historical search request and the historical recall advertisement, and shielding advertisements with correlation lower than a threshold value during advertisement recall;

the determining the target entity node associated with the historical search request and the historical recall advertisement in the target domain knowledge graph comprises the following steps:

selecting a target entity node from the candidate entity nodes according to the core word;

the core words are word segmentation words which can best embody the core content of the target text; the candidate entity nodes are entity nodes which can embody the core content of the target text in all entity nodes associated with word segmentation of the target domain knowledge graph hit by the target text; the target entity node is the entity node which can embody the core content of the target text most.

2. The method according to claim 1, wherein determining the core word of the target text according to the matching result between the word segmentation result of the target text and the nodes in the target domain knowledge-graph comprises:

3. The method according to claim 1, wherein determining candidate entity nodes of the target text according to a matching result between the word segmentation result of the target text and the nodes in the target domain knowledge-graph comprises:

4. The method of claim 1, further comprising, after said determining a blacklist advertisement based on a correlation between said historical search request and said historical recall advertisement:

5. The method of claim 1, wherein the target domain knowledge-graph is a medical domain knowledge-graph.

6. The method of claim 5, further comprising, prior to determining a target entity node with which the historical search request and the historical recall advertisement are associated in the target domain knowledge base, and a target attribute node connected to the target entity node:

7. The method of claim 6, wherein constructing a medical knowledge-graph with solid nodes as cores comprises:

8. The method of claim 5, wherein the determining a correlation between the historical search request and the historical recall advertisement according to the target attribute node comprises:

9. An advertising recall device, comprising:

the advertisement shielding module is used for determining blacklist advertisements according to the correlation between the historical search request and the historical recall advertisements and shielding advertisements with correlation lower than a threshold value during advertisement recall;

the node matching module is specifically configured to perform word segmentation processing on the historical search request and a target text in the historical recall advertisement;

selecting a target entity node from the candidate entity nodes according to the core word; the core words are word segmentation words which can best embody the core content of the target text; the candidate entity nodes are entity nodes which can embody the core content of the target text in all entity nodes associated with word segmentation of the target domain knowledge graph hit by the target text; the target entity node is the entity node which can embody the core content of the target text most.

10. An electronic device, comprising:

at least one processor; and

a memory communicatively coupled to the at least one processor; wherein,

the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the advertising recall method of any one of claims 1-8.

11. A non-transitory computer readable storage medium storing computer instructions for causing the computer to perform the advertising recall method of any one of claims 1-8.