CN113190593A - Search recommendation method based on digital human knowledge graph - Google Patents
Search recommendation method based on digital human knowledge graph Download PDFInfo
- Publication number
- CN113190593A CN113190593A CN202110515371.3A CN202110515371A CN113190593A CN 113190593 A CN113190593 A CN 113190593A CN 202110515371 A CN202110515371 A CN 202110515371A CN 113190593 A CN113190593 A CN 113190593A
- Authority
- CN
- China
- Prior art keywords
- entity
- user
- entities
- search
- digital human
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 28
- 238000012549 training Methods 0.000 claims abstract description 10
- 238000011160 research Methods 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000009792 diffusion process Methods 0.000 claims description 5
- 239000011159 matrix material Substances 0.000 claims description 5
- 230000006870 function Effects 0.000 claims description 4
- 235000008694 Humulus lupulus Nutrition 0.000 claims description 3
- 230000004044 response Effects 0.000 claims description 2
- 238000010845 search algorithm Methods 0.000 claims description 2
- 238000010276 construction Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000010365 information processing Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002457 bidirectional effect Effects 0.000 description 1
- 210000004204 blood vessel Anatomy 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2457—Query processing with adaptation to user needs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/28—Databases characterised by their database models, e.g. relational or object models
- G06F16/284—Relational databases
- G06F16/288—Entity relationship models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
- G06F40/295—Named entity recognition
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Audiology, Speech & Language Pathology (AREA)
- General Health & Medical Sciences (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a search recommendation method based on a digital human knowledge map, which comprises the following steps: constructing a digital human knowledge map; setting search attributes and creating indexes to search entities; recommending entities selected by the user to be similar entities of the same type; simulating an interest propagation path of the user on a knowledge graph according to RippleNet model training data recorded by the user, and taking an entity with a higher predicted value as a recommended entity of the user; and returning the recommended entities of the same type and entities predicted by the search records of the learning user to be possibly interested by the user to the user.
Description
Technical Field
The invention relates to the technical field of natural language and computer information processing, in particular to a search recommendation method based on a digital human knowledge map.
Background
The existing Chinese culture is the spirit blood vessels continued by Chinese nationalities for thousands of years, and the digital humanity combines the ancient Chinese culture with the modern information technology, provides a new method and a new thought for the culture history research, develops and develops the culture spirit with the contemporary value, and becomes the research trend of humanity science. The knowledge graph is used as a visualization tool for displaying the field of complex knowledge, and can provide high-quality and multi-dimensional reference information for subject research by combining technologies such as data mining and information processing, so that the knowledge graph is more and more widely applied to a plurality of fields. In the digital human system, when a user wants to know human related knowledge, other contents which the user may be interested in are recommended by means of a knowledge graph and a related recommendation algorithm while searching a target knowledge point of the user, so that the user can be helped to acquire more human knowledge, contact is established in the brain, impression is deepened, and the method is also helpful for professional researchers to expand thinking and find a new research direction.
Disclosure of Invention
In order to solve the technical problems, the invention aims to provide a searching and recommending method based on a digital human knowledge graph, which expounds the implementation processes of three stages of construction, searching and recommending of the digital human knowledge graph and provides an idea for the application of the digital human knowledge graph.
The purpose of the invention is realized by the following technical scheme:
a search recommendation method based on a digital human knowledge graph comprises the following steps:
step A, constructing a digital humanity knowledge map;
step B, setting search attributes, creating indexes and searching entities;
step C, recommending that the entities selected by the user are similar entities of the same type;
step D, simulating an interest propagation path of the user on a knowledge graph according to RippleNet model training data for the user search record, and taking an entity with a higher predicted value as a recommended entity of the user;
and E, returning the recommended entities of the same type and entities predicted by the search records of the learning user and possibly interested by the user to the user.
One or more embodiments of the present invention may have the following advantages over the prior art:
the method can provide reference for map construction, entity search and related entity recommendation in the digital human language field, and promote digital and intelligent development of historical culture research.
Drawings
FIG. 1 is a flow chart of a digital human knowledge graph-based search recommendation method;
FIGS. 2 and 3 are digital human figure entity data charts, respectively;
FIG. 4 is a digital human event entity data chart;
FIG. 5 is a diagram of the main Cypher sentences that implement the search.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the following embodiments and accompanying drawings.
The main technical core of the embodiment lies in that the types and the relations of main entities in the digital human language field are defined in a knowledge graph framework, the data source is reliable, and the constructed digital human language knowledge graph can meet the requirements of users; the entity searching process considers synonym searching, combines accurate searching and fuzzy searching and has higher searching efficiency; the entity recommendation combines two aspects of similar entity recommendation of the same type and entity recommendation which is analyzed by user historical search records and is possibly interested by a user, the diversity of recommendation results is enriched, and the two recommendations are respectively realized by path analysis and a RippleNet algorithm.
As shown in fig. 1, a process of a search recommendation method based on a digital human knowledge graph includes the following steps:
firstly, defining a framework of a digital human knowledge graph, wherein the framework comprises entities and relationship types, analyzing the research direction of the digital human knowledge graph, dividing the entities related to the graph into five categories, namely, characters, events, regions, officials, dynasties and the like, wherein bidirectional relationships exist among the entities of different categories, relationship names are defined as the types of relationship pointing to the entities, particularly, the characters and the entities also have relationships in front, and the relationship names are mainly classified into a relationship and a non-relationship. As shown in table 1, the entity structure table of the digital human knowledge map is:
some relationship attributes defined in this embodiment are as follows, and in addition to the following table, relationships may also exist between people and entities, and the main categories are related to relatives and social relationships as shown in table 2:
TABLE 2
Depending on data such as a Character Biography Database (CBDB) of the original data and entries of two million tool books recorded by the Hopkins, required information is extracted, five entity tables and entity relation tables are obtained through sorting and stored in a MySQL database, and the five entity tables, such as a character entity table, an event entity table and a character event relation table, are shown in FIGS. 2, 3 and 4.
Constructing a digital human knowledge graph by using a Neo4j high-performance NoSQL graph database according to the graph framework and the data; the collated entity and relationship data are used for constructing a digital human knowledge graph by Neo4j, and about 63 ten thousand entity nodes and 287 ten thousand relationships exist.
in historical data, people, events, officials and the like have aliases, the aliases are considered to bring better use experience to users when searching entities, but the aliases of the people may comprise a plurality of attribute fields such as characters, numbers, envelope numbers, and secondary numbers, and an index needs to be created for the fields; and the entity quantity is huge, and if the search result is expanded by fuzzy search, the search performance is greatly reduced.
Based on the characteristics, the index creating method is designed as follows:
1) combining entity words and synonyms of entity words (including characters, numbers, aliases of search characters, alias of events and the like) into a new entity attribute for searching, namely combining an entity name and all aliases of an entity into a field for searching, and using separators; "separate.
2) The display index is established in Neo4j for the search field, the problem that the mode index does not support fuzzy search and full-text search is solved, the nodes can be searched by using text search in Neo4j, the grammar rule is the same as Lucene, and the search efficiency of the nodes is obviously improved.
3) And during searching, adopting the reference search grammar rule to preferentially and accurately search, and listing the search results for the user to select if the search results cannot be fuzzily matched.
The Cypher statement that creates an index and searches using the index (taking "li-white" as an example) is shown in fig. 5.
after a user selects a specific entity (hereinafter referred to as a central entity), a recommendation algorithm firstly recommends entities of the same type as the central entity, and can satisfy the relevance discovery research aiming at a certain type of entities, and the recommendation algorithm mainly comprises the following steps:
1) acquiring entities of the same type within k hops away from the selected entity, wherein the value of k is usually 3, namely, the entities of the same type within k hops away from the central entity in the knowledge graph are selected as candidate entities;
2) analyzing the paths of the entities and the selected entity to calculate the degree of association between the two entities, finding out the entities with short path length and large number of paths with the selected entity, namely, adopting a path search algorithm, analyzing the possible paths of each candidate entity and the central entity in the map, and calculating the association between the candidate entity and the central entity, wherein the calculation formula is as follows: :
wherein n is the total number of paths, m is the number of nodes except the source node and the target node on the ith path, and wijThe weight value of the jth node in the ith path is related to the type of the node, and specific values are given in table 1. According to the formula, the shorter the path between two nodes is, the more the number of paths is, the greater the relevance of the nodes is;
3) according to the relevance calculated in the step 2), recommending the entities with higher relevance as the entities of the same type which are possibly interested by the user.
Wherein, the larger the value of k is, the more associated entities will be dispersed, but the complexity of calculation will be synchronously increased, and generally 3 is taken.
the recommended entities do not limit the entity types, the diversity of recommended items is enriched, and the method is beneficial to finding a certain special research which is interested by the user. The RippleNet algorithm is adopted for recommending in the step, the interest propagation process of the user is simulated by ripple in the algorithm, and the problems that the traditional collaborative filtering recommendation algorithm cannot solve such as cold start and sparse relation matrix can be well solved. The framework of the rippenet algorithm is set forth below:
1) constructing an initial entity set by using a search record of a user, generating training data, namely collecting all searched entity items of the user as a seed set, and constructing the training data on the basis of the seed, wherein the training data comprises a positive case (item in the seed, the probability is 1) and a negative case (item outside the seed is randomly selected, and the probability is 0);
2) initializing an embedding layer of entity item and relation;
3) expanding an initial entity Set in a knowledge graph by one hop outwards on the knowledge graph (the maximum length needs to be Set to avoid overlarge data), constructing a primary interest propagation Ripple Set of a user, and expressing a triple in a network by (h, r, t) to obtain a result of the first round of diffusion of the user interest;
the normalized similarity of entity itemv to (h, r) is calculated using the following formula:
wherein,
set of 1-hop ripples of v, RiAnd hiAre respectively the relationship riAnd head entity hiBy the similarity piThe t is weighted and summed, as calculated below, to obtain the output o, vector of this layer
One stage of the user's historical click response to itemv can be seen.
4) Taking t of the Ripple Set of the first layer as h of the second layer, taking out the Ripple Set of the second layer, repeating the process of the previous stage, and taking the weighted sum of the similarity of the Ripple Set of the second layer as output
5) After repeating the process for H times, accumulating the calculation results of each round to obtain an embedded vector capable of expressing the interest of the user, calculating the inner product with the entity item vector, and then normalizing to calculate the click probability of the user on the item; the final user-embedded vector expression formula is:
at this time, the predicted value of the user click rate of entity itemv may be calculated as:
the above is the main framework of the RippleNet model, and the optimized loss function during model training is defined below, including the real y and RippleNet predicted values
Cross entropy loss between, error of reconstructed index matrix and real data, and regular term:
in the formula, a head entity h and a tail entity t are uniformly represented by an E matrix.
And continuously and iteratively updating Embedding, minimizing the loss function, finally obtaining a rippeNet model for predicting the user interest, predicting, and outputting an entity with a higher predicted value as a recommended entity of the user.
And continuously optimizing model parameters by considering loss functions such as cross entropy loss, regular terms and the like of the predicted value and the true value, finally obtaining a rippleNet model, predicting, and outputting an entity with a higher predicted value as a recommended entity of the user.
And 5, returning the recommended entities of the same type and entities predicted by the search records of the learning user to the user and possibly interested by the user.
Although the embodiments of the present invention have been described above, the above descriptions are only for the convenience of understanding the present invention, and are not intended to limit the present invention. It will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (7)
1. A search recommendation method based on a digital human knowledge graph is characterized by comprising the following steps:
step A, constructing a digital humanity knowledge map;
step B, setting search attributes, creating indexes and searching entities;
step C, recommending that the entities selected by the user are similar entities of the same type;
step D, simulating an interest propagation path of the user on a knowledge graph according to RippleNet model training data for the user search record, and taking an entity with a higher predicted value as a recommended entity of the user;
and E, returning the recommended entities of the same type and entities predicted by the search records of the learning user and possibly interested by the user to the user.
2. The digital human knowledge graph-based search recommendation method according to claim 1, wherein the step a comprises:
defining a digital human knowledge map framework, analyzing the research direction of the digital human field, and classifying entities related to the map;
extracting required information through database data, sorting to obtain a classified entity table and an entity relation table, and storing the classified entity table and the entity relation table into a database;
and constructing the digital human knowledge map through the graphic database according to the map frame and the data.
3. The digital human knowledge graph-based search recommendation method of claim 2, wherein the entities involved in the graph include five major categories of people, events, regions, official posts and dynasties.
4. The method for recommending searches based on digital human knowledge-graph according to claim 1, wherein in said step C, the same type of entity as the entity selected by the user, i.e. the central entity, is recommended by the recommendation algorithm; the recommendation algorithm specifically comprises:
1) selecting entities of the same type with the distance from a central entity being less than or equal to k hops in the knowledge graph as candidate entities;
2) adopting a path search algorithm, analyzing the possible paths of each candidate entity and the central entity in the map, and calculating the relevance between the candidate entities and the central entity, wherein the calculation formula is as follows:
wherein n is the total number of paths, m is the number of nodes except the source node and the target node on the ith path, and wijThe weighted value of the jth node in the ith path is related to the type of the node, and the formula shows that the shorter the path between the two nodes is, the more the number of the paths is, and the greater the relevance of the nodes is;
3) according to the relevance calculated in the step 2), recommending the entities with higher relevance as the entities of the same type which are possibly interested by the user.
5. The digital human knowledge graph-based search recommendation method according to claim 1, wherein the step D specifically comprises:
1) constructing an initial entity set from the search records of the user to generate training data;
2) initializing an embedded layer in an algorithm module;
3) expanding the initial entity set outwards by one layer in the knowledge graph, calculating the similarity of an entity item with the initial entity and the relation inner product, and performing weighted summation on the similarity and the target entity to obtain a result of the user interest after the first round of diffusion;
4) taking the tail node of the first round as a head node, expanding one hop outwards, and obtaining an interesting second round diffusion result according to the same method;
5) after diffusion for many times, accumulating the calculation results of each round to obtain an embedded vector capable of expressing the interest of the user, calculating an inner product with the entity item vector, and then normalizing to calculate the click probability of the user on the item.
6. The method of claim 5, wherein the similarity in 3) is obtained by calculating the similarity of the entity item v and the inner product of the initial entity and the relationship (h, r) in the map triple and normalizing, and the calculation formula is as follows:
wherein v is an embedding vector of the entity item of the click rate to be predicted,
set of 1-hop ripples of v, RiAnd hiAre respectively the relationship riAnd head entity hiBy the similarity piThe t is weighted and summed, as calculated below, to obtain the output o, vector of this layer
One can see a one-phase response of the user's historical clicks to itemv, where:
7. the digital human knowledge graph-based search recommendation method according to claim 5, wherein the 5) specifically comprises: repeating the processes from 1) to 4) for H times, and adding the representations obtained by multiple interest diffusion to obtain a final user embedded vector representation:
at this time, the predicted value of the user click rate of entity itemv may be calculated as:
the above is the main framework of the RippleNet model, and the optimized loss function during model training is defined below, including the real y and RippleNet predicted values
Cross entropy loss between, error of reconstructed index matrix and real data, and regular term:
in the formula, a head entity h and a tail entity t are uniformly represented by an E matrix.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110515371.3A CN113190593A (en) | 2021-05-12 | 2021-05-12 | Search recommendation method based on digital human knowledge graph |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110515371.3A CN113190593A (en) | 2021-05-12 | 2021-05-12 | Search recommendation method based on digital human knowledge graph |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113190593A true CN113190593A (en) | 2021-07-30 |
Family
ID=76981271
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110515371.3A Pending CN113190593A (en) | 2021-05-12 | 2021-05-12 | Search recommendation method based on digital human knowledge graph |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113190593A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114186075A (en) * | 2021-12-14 | 2022-03-15 | 武汉数文科技有限公司 | Semantic search method for knowledge graph in culture field |
| CN115329200A (en) * | 2022-08-26 | 2022-11-11 | 国家开放大学 | Teaching resource recommendation method based on knowledge graph and user similarity |
| CN115618108A (en) * | 2022-10-24 | 2023-01-17 | 云南中烟工业有限责任公司 | Cigarette brand recommendation method based on knowledge graph in new retail mode |
| CN116304303A (en) * | 2023-02-01 | 2023-06-23 | 北京三维天地科技股份有限公司 | Asset recommendation method and system based on knowledge graph |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109710701A (en) * | 2018-12-14 | 2019-05-03 | 浪潮软件股份有限公司 | A kind of automated construction method for public safety field big data knowledge mapping |
| CN110110094A (en) * | 2019-04-22 | 2019-08-09 | 华侨大学 | Across a network personage's correlating method based on social networks knowledge mapping |
| CN110188208A (en) * | 2019-06-04 | 2019-08-30 | 河海大学 | A kind of the information resources inquiry recommended method and system of knowledge based map |
| CN110968699A (en) * | 2019-11-01 | 2020-04-07 | 数地科技(北京)有限公司 | Logic map construction and early warning method and device based on event recommendation |
| CN111259133A (en) * | 2020-01-17 | 2020-06-09 | 成都信息工程大学 | Personalized recommendation method integrating multiple information |
| CN111523029A (en) * | 2020-04-20 | 2020-08-11 | 浙江大学 | Personalized recommendation method based on knowledge graph representation learning |
| CN111986006A (en) * | 2020-09-02 | 2020-11-24 | 深圳壹账通智能科技有限公司 | Product recommendation method and device based on knowledge graph, computer equipment and storage medium |
| CN112131490A (en) * | 2020-09-18 | 2020-12-25 | 东南大学 | Region-sensitive interest point recommendation method driven by knowledge graph |
| CN112214685A (en) * | 2020-09-27 | 2021-01-12 | 电子科技大学 | Knowledge graph-based personalized recommendation method |
| CN112612974A (en) * | 2021-01-04 | 2021-04-06 | 上海明略人工智能(集团)有限公司 | Friend recommendation method and system based on path sorting |
| CN112712056A (en) * | 2021-01-14 | 2021-04-27 | 珠海格力电器股份有限公司 | Video semantic analysis method and device, storage medium and electronic equipment |
| CN112732932A (en) * | 2021-01-08 | 2021-04-30 | 西安烽火软件科技有限公司 | User entity group recommendation method based on knowledge graph embedding |
| CN112765322A (en) * | 2021-01-25 | 2021-05-07 | 河海大学 | Remote sensing image search recommendation method based on water conservancy domain knowledge graph |
| CN112765486A (en) * | 2021-01-22 | 2021-05-07 | 重庆邮电大学 | Knowledge graph fused attention mechanism movie recommendation method |
-
2021
- 2021-05-12 CN CN202110515371.3A patent/CN113190593A/en active Pending
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109710701A (en) * | 2018-12-14 | 2019-05-03 | 浪潮软件股份有限公司 | A kind of automated construction method for public safety field big data knowledge mapping |
| CN110110094A (en) * | 2019-04-22 | 2019-08-09 | 华侨大学 | Across a network personage's correlating method based on social networks knowledge mapping |
| CN110188208A (en) * | 2019-06-04 | 2019-08-30 | 河海大学 | A kind of the information resources inquiry recommended method and system of knowledge based map |
| CN110968699A (en) * | 2019-11-01 | 2020-04-07 | 数地科技(北京)有限公司 | Logic map construction and early warning method and device based on event recommendation |
| CN111259133A (en) * | 2020-01-17 | 2020-06-09 | 成都信息工程大学 | Personalized recommendation method integrating multiple information |
| CN111523029A (en) * | 2020-04-20 | 2020-08-11 | 浙江大学 | Personalized recommendation method based on knowledge graph representation learning |
| CN111986006A (en) * | 2020-09-02 | 2020-11-24 | 深圳壹账通智能科技有限公司 | Product recommendation method and device based on knowledge graph, computer equipment and storage medium |
| CN112131490A (en) * | 2020-09-18 | 2020-12-25 | 东南大学 | Region-sensitive interest point recommendation method driven by knowledge graph |
| CN112214685A (en) * | 2020-09-27 | 2021-01-12 | 电子科技大学 | Knowledge graph-based personalized recommendation method |
| CN112612974A (en) * | 2021-01-04 | 2021-04-06 | 上海明略人工智能(集团)有限公司 | Friend recommendation method and system based on path sorting |
| CN112732932A (en) * | 2021-01-08 | 2021-04-30 | 西安烽火软件科技有限公司 | User entity group recommendation method based on knowledge graph embedding |
| CN112712056A (en) * | 2021-01-14 | 2021-04-27 | 珠海格力电器股份有限公司 | Video semantic analysis method and device, storage medium and electronic equipment |
| CN112765486A (en) * | 2021-01-22 | 2021-05-07 | 重庆邮电大学 | Knowledge graph fused attention mechanism movie recommendation method |
| CN112765322A (en) * | 2021-01-25 | 2021-05-07 | 河海大学 | Remote sensing image search recommendation method based on water conservancy domain knowledge graph |
Non-Patent Citations (1)
| Title |
|---|
| 慕佳芯: "基于知识图谱和偏好传播的电影推荐算法", 《中国优秀硕士论文全文数据库》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114186075A (en) * | 2021-12-14 | 2022-03-15 | 武汉数文科技有限公司 | Semantic search method for knowledge graph in culture field |
| CN114186075B (en) * | 2021-12-14 | 2024-04-26 | 武汉数文科技有限公司 | Semantic search method for knowledge graph in cultural domain |
| CN115329200A (en) * | 2022-08-26 | 2022-11-11 | 国家开放大学 | Teaching resource recommendation method based on knowledge graph and user similarity |
| CN115329200B (en) * | 2022-08-26 | 2024-04-26 | 国家开放大学 | Teaching resource recommendation method based on knowledge graph and user similarity |
| CN115618108A (en) * | 2022-10-24 | 2023-01-17 | 云南中烟工业有限责任公司 | Cigarette brand recommendation method based on knowledge graph in new retail mode |
| CN115618108B (en) * | 2022-10-24 | 2023-05-26 | 云南中烟工业有限责任公司 | Brand recommendation method based on knowledge graph in new retail model |
| CN116304303A (en) * | 2023-02-01 | 2023-06-23 | 北京三维天地科技股份有限公司 | Asset recommendation method and system based on knowledge graph |
| CN116304303B (en) * | 2023-02-01 | 2023-09-08 | 北京三维天地科技股份有限公司 | Asset recommendation method and system based on knowledge graph |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108052583B (en) | E-commerce ontology construction method | |
| US10977581B2 (en) | Systems and methods for secondary knowledge utilization in machine learning | |
| WO2020108608A1 (en) | Search result processing method, device, terminal, electronic device, and storage medium | |
| CN107341145B (en) | A kind of user feeling analysis method based on deep learning | |
| CN113190593A (en) | Search recommendation method based on digital human knowledge graph | |
| US8204988B2 (en) | Content-based and time-evolving social network analysis | |
| US12013850B2 (en) | Method and system for advanced data conversations | |
| Gilson et al. | From web data to visualization via ontology mapping | |
| Devi et al. | ADANS: An agriculture domain question answering system using ontologies | |
| CN104573130B (en) | The entity resolution method and device calculated based on colony | |
| CN106383836B (en) | Attributing actionable attributes to data describing an identity of an individual | |
| CN108664509B (en) | Method, device and server for ad hoc query | |
| WO2020010834A1 (en) | Faq question and answer library generalization method, apparatus, and device | |
| CN112966091B (en) | Knowledge map recommendation system fusing entity information and heat | |
| JP2009093649A (en) | Recommendation for term specifying ontology space | |
| Ignatov et al. | Can triconcepts become triclusters? | |
| CN111460102B (en) | Chart recommendation system and method based on natural language processing | |
| CN110083683B (en) | Entity semantic annotation method based on random walk | |
| KR20140067697A (en) | System and method for supplying collaboration partner search service | |
| CN106708929A (en) | Video program searching method and device | |
| CN112000790B (en) | Legal text accurate retrieval method, terminal system and readable storage medium | |
| CN110110218A (en) | A kind of Identity Association method and terminal | |
| CN112199958A (en) | Concept word sequence generation method and device, computer equipment and storage medium | |
| CN103942232A (en) | Method and equipment for mining intentions | |
| CN113987126A (en) | Retrieval method and device based on knowledge graph |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |