CN112948568B - Content recommendation method and device based on text concept network - Google Patents
Content recommendation method and device based on text concept network Download PDFInfo
- Publication number
- CN112948568B CN112948568B CN201911258023.1A CN201911258023A CN112948568B CN 112948568 B CN112948568 B CN 112948568B CN 201911258023 A CN201911258023 A CN 201911258023A CN 112948568 B CN112948568 B CN 112948568B
- Authority
- CN
- China
- Prior art keywords
- text
- concept
- recommended
- network
- user
- 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.)
- Active
Links
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
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/335—Filtering based on additional data, e.g. user or group profiles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a content recommendation method and a content recommendation device based on a text concept network, wherein the method firstly utilizes m independent texts to construct a concept library, then constructing an initial concept set of the user according to the public concepts screened out by the concept library and basic concepts extracted from the basic education textbook, and finally, determining whether the text to be recommended is recommended according to the matching and linking conditions of concept nodes contained in the text concept network corresponding to the text to be recommended in the user concept network. The method of the invention can improve the recommendation effect and is beneficial to expanding the interested field of the user.
Description
Technical Field
The invention relates to the technical field of content recommendation, in particular to a content recommendation method and device based on a text concept network.
Background
The recommendation system is an important technical link for interaction between internet data and users. In the prior art, a common content recommendation method generally classifies contents to be recommended based on manual labels or vocabulary analysis included in the contents, and recommends related classified contents according to the field of interest of a user.
In the process of implementing the present invention, the inventors of the present application find that the methods in the prior art have at least the following technical problems:
due to the complexity of the content information, errors often occur in labels and classification, meanwhile, the recommendation which simply depends on the content classification is difficult to expand the interest range of the user, and the unknown interest field of the user can be tried only by inserting some random hot content. Therefore, the recommendation is uncontrollable, the obtained user data is not structural and representative, and the improvement and the expansion of the user portrait are difficult to be better served.
Therefore, the method in the prior art has the technical problems that the recommendation effect is poor, and the interest range of the user is difficult to expand.
Disclosure of Invention
In view of the above, the present invention provides a content recommendation method and apparatus based on a text concept network, so as to solve or at least partially solve the technical problems of poor recommendation effect and difficulty in expanding the user's interest range in the prior art.
In order to solve the above technical problem, a first aspect of the present invention provides a content recommendation method based on a text concept network, including:
analyzing each text contained in a text set containing m independent texts, and constructing a concept library, wherein the concept library comprises concepts, and m is a positive integer;
screening out public concepts according to the proportion of concepts in the concept library appearing in m independent texts, analyzing a basic education teaching material, extracting basic concepts, combining the basic concepts and the public concepts into a user initial concept set, and constructing an initialization network according to the matching condition of texts in the basic education teaching material and the concepts in the user initial concept set, wherein nodes of the initialization network are used for representing the concepts in the basic education teaching material matched with the user initial concept set, and edges of the initialization network represent links between the concepts matched with the user initial concept set;
acquiring a user text, and constructing a user text network according to the matching condition of the user text and a concept library, wherein nodes in the user text network are used for representing concepts matched with the concept library in the user text, and edges in the user text network represent links between the concepts matched with the concept library;
merging the initialization network and the user text network into a user concept network;
constructing a text concept network corresponding to each text to be recommended according to the matching condition of the text to be recommended in the text library to be recommended and the concept library, wherein nodes of the text concept network corresponding to the text to be recommended are used for representing concepts matched with the concept library in the text to be recommended, and edges of the text concept network corresponding to the text to be recommended represent links between the concepts matched with the concept library;
constructing a total concept network according to the text concept network corresponding to each text to be recommended, wherein nodes in the total concept network are used for representing concepts in all the text concept networks, and edges in the total concept network are used for representing links among the concepts contained in the text concept networks;
and determining whether to recommend the text to be recommended according to the matching condition of the concept nodes contained in the text concept network corresponding to the text to be recommended in the total concept network and the matching and linking condition of the concept nodes contained in the text concept network corresponding to the text to be recommended in the user concept network.
In one embodiment, parsing each text included in a text collection including m independent texts to construct a concept library includes:
preprocessing each text contained in a text set containing m independent texts to obtain all sentences and vocabularies of a corpus formed by the m independent texts, wherein m is a positive integer;
taking each vocabulary of the corpus as a concept subject word, traversing all sentences and vocabularies, and bringing the vocabulary which commonly appears in the same sentence with the concept subject word into a vocabulary set corresponding to the concept subject word, wherein the vocabulary set comprises the concept subject word and vocabulary elements;
and screening vocabulary elements of each vocabulary set to construct a concept library.
In one embodiment, the vocabulary element screening is performed on each vocabulary set, and a concept library is constructed, wherein the method comprises the following steps:
counting each vocabulary element x in the vocabulary set j And concept topic word x i The number z of the texts which appear together, wherein z is less than or equal to m;
and judging whether the text quantity z is larger than or equal to a first threshold value, if so, taking the vocabulary elements as effective vocabularies of the vocabulary set and keeping the effective vocabularies in the vocabulary set, otherwise, removing the vocabulary elements from the vocabulary set.
In one embodiment, constructing a user text network according to matching of user text with a concept library comprises:
judging whether the user text contains concepts in a concept library;
and taking the concepts contained in the user text as nodes, and taking a connecting line between every two concepts as an edge to construct a user text network.
In one embodiment, determining whether to recommend a text to be recommended according to matching and linking conditions of concept nodes contained in a text concept network corresponding to the text to be recommended in a user concept network includes:
if all concept nodes contained in the text concept network corresponding to the text to be recommended are matched with nodes in the user concept network, and all concept nodes contained in the text concept network corresponding to the text to be recommended are fully connected in the user concept network, the text to be recommended is not recommended, the text to be recommended is deleted from the text library to be recommended, a candidate text library is obtained, and otherwise, the text to be recommended is recommended.
In one embodiment, recommending a text to be recommended includes:
acquiring incompleteness of the text to be recommended and importance of missing content according to missing conditions of links and nodes in a text concept network corresponding to the text to be recommended;
and orderly recommending the texts to be recommended in the candidate text library according to the incompleteness of the texts to be recommended and the importance of the missing content.
In one embodiment, obtaining the incompleteness of the text to be recommended and the importance of the missing content according to the missing condition of links and nodes in the text concept network corresponding to the text to be recommended includes:
determining the incompleteness of the text to be recommended according to the number of the missing nodes and the number of the missing links;
determining the importance of the missing content according to the criticality of the missing link and the criticality of the missing node in the text concept network corresponding to the text to be recommended, wherein the criticality of the missing link is determined by whether two concept nodes connected by the missing link belong to the same category, and the criticality of the missing node is determined according to the degree of the missing node in the total concept network.
In one embodiment, according to the incompleteness of the text to be recommended and the importance of the missing content, orderly recommending the text to be recommended in a candidate text library, including:
quantifying the incompleteness of the text to be recommended and the importance of the missing content to respectively obtain corresponding calculation formulas, wherein the incompleteness calculation formula of the text to be recommended is as follows: s Δ ＝log 2 (N Δlink )+N Δnode The formula for calculating the importance of the missing content is I Δ ＝N T(link) +D Δnode ，N Δlink Indicates the number of missing links, N Δnode Indicates the number of missing nodes, N T(link) Denotes the sum of the criticalities of all missing links, D Δnode Representing the degree of all missing nodes in the overall concept network;
will S Δ And I Δ Normalizing to obtain two numerical dimensions S of text recommendation Δ And I Δ And according to two numerical dimensions SΔAnd IΔCalculating a recommended text recommendation score S A Wherein, in the process,
according to S A The text in the candidate text library is recommended in order according to the numerical value of the text.
Based on the same inventive concept, a second aspect of the present invention provides a content recommendation apparatus based on a text concept network, comprising:
the concept library construction module is used for analyzing each text contained in a text set containing m independent texts and constructing a concept library, wherein the concept library comprises concepts, and m is a positive integer;
the system comprises an initialization network construction module, a basic education teaching material and an initialization network, wherein the initialization network construction module is used for screening out public concepts according to the proportion of concepts in a concept library appearing in m independent texts, analyzing the basic education teaching material and extracting basic concepts, merging the basic concepts and the public concepts into a user initial concept set, and constructing the initialization network according to the matching condition of texts in the basic education teaching material and the concepts in the user initial concept set, wherein nodes of the initialization network are used for representing the concepts in the basic education teaching material, which are matched with the user initial concept set, and edges of the initialization network represent links between the concepts matched with the user initial concept set;
the user text network construction module is used for acquiring a user text and constructing a user text network according to the matching condition of the user text and the concept library, wherein nodes in the user text network are used for representing concepts matched with the concept library in the user text, and edges in the user text network represent links between the concepts matched with the concept library;
the user concept network construction module is used for combining the initialization network and the user text network into a user concept network;
the text concept network construction module is used for constructing a text concept network corresponding to each text to be recommended according to the matching condition of the text to be recommended in the text library to be recommended and the concept library, wherein the nodes of the text concept network corresponding to the text to be recommended are used for representing the concepts matched with the concept library in the text to be recommended, and the edges of the text concept network corresponding to the text to be recommended represent the links between the concepts matched with the concept library;
the general concept network construction module is used for constructing a general concept network according to the text concept network corresponding to each text to be recommended, nodes in the general concept network are used for representing concepts in all the text concept networks, and edges in the general concept network are used for representing links among the concepts;
and the recommending module is used for determining whether to recommend the text to be recommended according to the matching condition of the concept nodes contained in the text concept network corresponding to the text to be recommended in the total concept network and the matching and linking condition of the concept nodes contained in the text concept network corresponding to the text to be recommended in the user concept network.
Based on the same inventive concept, a third aspect of the present invention provides a computer-readable storage medium having stored thereon a computer program which, when executed, performs the method of the first aspect.
One or more technical solutions in the embodiments of the present application have at least one or more of the following technical effects:
the invention provides a content recommendation method based on a text concept network, which comprises the steps of firstly constructing a concept library by using m independent texts, then constructing a user initial concept set according to public concepts screened out by the concept library and basic concepts extracted from basic education teaching materials, further constructing an initialization network, then constructing a user text network according to the matching condition of a user text and the concept library, merging the initialization network and the user text network into a user concept network, then constructing a text concept network corresponding to each text to be recommended according to the matching condition of the text to be recommended in the text library to be recommended and the concept library, then constructing a total concept network according to the text concept network corresponding to each text to be recommended, and finally constructing a total concept network according to the matching condition of concept nodes contained in the text concept network corresponding to the text to be recommended in the total concept network, And determining whether the text to be recommended is recommended according to the matching and linking conditions of concept nodes contained in the text concept network corresponding to the text to be recommended in the user concept network.
Because the method provided by the invention can carry out path query on the text concept network corresponding to the text to be recommended based on the user concept network node, analyze the matching and linking conditions of the concept nodes contained in the text concept network corresponding to the text to be recommended in the user concept network, thereby realizing ordered recommendation, compared with the existing method for recommending content based on labels and vocabularies contained in the text to be recommended, the method characterizes the concept by the collection of vocabularies related to the concept vocabularies, and respectively constructs the user text network according to the matching conditions of the text in the text library and the concept library of the user, the text concept network can utilize the text concept network structure to gradually expand or strengthen the structure of the text concept network of the user based on the content mastered by the user according to the characteristics of the text concept network corresponding to the text to be recommended, thereby realizing the ordered recommendation of the content.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
Fig. 1 is a schematic flowchart of a content recommendation method based on a text concept network according to an embodiment of the present invention;
fig. 2 is a block diagram illustrating a content recommendation apparatus based on a text concept network according to an embodiment of the present invention;
fig. 3 is a block diagram of a computer-readable storage medium according to an embodiment of the present invention.
Detailed Description
The invention aims to provide a content recommendation method and device based on a text concept network aiming at the technical problems that the recommendation effect is poor and the interest range of a user is difficult to expand in the method in the prior art, so that the purposes of improving the recommendation effect and expanding the interest range of the user are achieved.
In order to achieve the above object, the main concept of the present invention is as follows:
the construction of a text concept network is based on the same concepts that are present in different texts, and concepts are characterized by means of a collection of words related to the concept words. The relevance of the words in the word set is constructed according to the co-occurrence rule of the words in the text. According to the rules, a user text network and a text concept network are respectively constructed for related data of a user and a text database to be recommended, and based on a node of the user concept network formed by combining an initialization network and the user text network, path query is carried out on the text concept network corresponding to the text to be recommended, so that content recommendation is realized.
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Example one
The embodiment provides a content recommendation method based on a text concept network, please refer to fig. 1, and the method includes:
step S1: analyzing each text contained in a text set containing m independent texts, and constructing a concept library, wherein the concept library comprises concepts, and m is a positive integer.
Specifically, the number of m may be determined according to actual conditions.
In one embodiment, parsing each text included in a text collection including m independent texts to construct a concept library includes:
preprocessing each text contained in a text set containing m independent texts to obtain all sentences and vocabularies of a corpus formed by the m independent texts, wherein m is a positive integer;
taking each vocabulary of the corpus as a concept subject word, traversing all sentences and vocabularies, and bringing the vocabulary which commonly appears in the same sentence with the concept subject word into a vocabulary set corresponding to the concept subject word, wherein the vocabulary set comprises the concept subject word and vocabulary elements;
and screening vocabulary elements of each vocabulary set to construct a concept library.
Specifically, after all the words of the corpus are obtained, on the basis of each word, the words which are in the same sentence with the word are searched, a word set is constructed by the words which commonly appear with the word, and the words which commonly appear with the word represent that the two words have association, wherein the word set comprises two words, one is a concept subject word and the other is a word element.
In one embodiment, the vocabulary element screening is performed on each vocabulary set, and a concept library is constructed, wherein the method comprises the following steps:
counting each vocabulary element x in the vocabulary set j And concept topic word x i The number z of the texts which appear together, wherein z is less than or equal to m;
and judging whether the text quantity z is larger than or equal to a first threshold value, if so, taking the vocabulary elements as effective vocabularies of the vocabulary set and keeping the effective vocabularies in the vocabulary set, otherwise, removing the vocabulary elements from the vocabulary set.
Specifically, to improve the accuracy of the concept, the step further filters the vocabulary element, for example, by determining whether the vocabulary element frequently appears in a text together with the concept topic word to determine whether the vocabulary element remains, wherein the frequent occurrence may be determined according to a set threshold.
Step S2: screening out public concepts according to the proportion of the concepts in the concept library appearing in m independent texts, analyzing a basic education teaching material, extracting the basic concepts, combining the basic concepts and the public concepts into a user initial concept set, and constructing an initialization network according to the matching condition of the texts in the basic education teaching material and the concepts in the user initial concept set, wherein the nodes of the initialization network are used for representing the concepts matched with the user initial concept set in the basic education teaching material, and the edges of the initialization network represent links between the concepts matched with the user initial concept set.
In particular, common concepts refer to concepts that occur in most texts (occurrence rate thresholds may be set, e.g., 60%, 70%, 80%, etc., depending on the situation), which may be "in the hands of" a majority of people, so the present invention takes them as part of initializing the network. Basic educational textbooks are another part of the text set used to construct the initialized network, because almost all people have accepted basic education, the concepts present in these textbooks are also the concepts "mastered" by most people and thus also added to the initialized network.
The purpose of presetting the initialization network is to provide a preset network for a user, so that the user has a recommended basis before reading or writing data. That is, the initialization network is an initial state of each user when no data is generated, and is used for solving the problem of cold start of the recommendation system, and purposeful recommendation can be realized without user data.
In the specific implementation, the concepts in the concept library analyzed in S1 can be sorted, for example, a concept X g Appear in more than 60% of the text, then concept X will be g As a common concept.
Step S3: the method comprises the steps of obtaining a user text, and constructing a user text network according to the matching condition of the user text and a concept library, wherein nodes in the user text network are used for representing concepts matched with the concept library in the user text, and edges in the user text network represent links between the concepts matched with the concept library.
Specifically, the user text refers to the text content read or input by all users, i.e., the historical data of the users.
In one embodiment, constructing a user text network according to matching of user text with a concept library comprises:
judging whether the user text contains concepts in a concept library;
and taking the concepts contained in the user text as nodes, and taking a connecting line between every two concepts as an edge to construct a user text network.
Specifically, the independent text A to be recommended is divided into sentences and words, stop words are removed, and a is obtained 1 ，a 2 ...a i I words are total. And statistically screening the concepts contained in the user text according to the constructed library constructed by the S1.
For example, when a vocabulary group (concept) X I In which 30% of the necessary vocabularies appear in the text w i In (1), the text w is determined i Containing vocabulary group X I . If the vocabulary group X I If the number of necessary words in the vocabulary is less than 5, the necessary words and the effective words are merged and calculated. When X is present I If the total number of vocabulary elements is less than 3, the text w is determined as long as 1 element appears i Containing the vocabulary group X I . When X is present I If the set is empty, then only the subject word x is required i If it appears, the text w is determined i Containing the vocabulary group X I . Further screening of the matched concepts, e.g. comparing two concepts X 1 And X 2 (the subject term is x respectively 1 And x 2 ) If X is 1 And X 2 If the selected effective vocabulary sets are the same, comparing the subject words x 1 And x 2 If the subject word x of one of the concepts 1 Is another concept X 2 The necessary words of (1) then only retain the concept X 2 In other cases both concepts remain.
The necessary vocabulary acquisition step may be:
will concept subject word x i The number of texts appearing is calculated as y (y is less than or equal to m) and the vocabulary element x j With the central subject term x i Co-occurrence frequency F between x Is equal to
The construction of the initial word vector is described below, and the training method of the word vector is as follows: for a text material set W containing m independent texts ═ W 1 ，w 2 ...w m Dividing each text into sentences and words, and removing stop words to obtain x 1 ，x 2 ...x n N words in total. Word embedding is carried out by using a word2vec method to obtain word vectors of n vocabularies, and the dimensionality of the word vectors is 200.
Each valid vocabulary x according to the word vectors obtained in advance j Distance concept topic word x i Is recorded as D x . With each collection vocabulary element x i F of (A) x And D x Two are providedIn a dimension-formed rectangular coordinate system, a vocabulary element x i The distance from the text coordinate to the origin is recorded as a lexical importance parameter I x ，
Get I x The vocabulary of the first 20% of the numerical value is used as the necessary vocabulary.
The following describes the construction of a user text network, and with the foregoing method, it can be determined whether a text contains a certain concept, and at this time, a network is constructed according to all concepts contained in a text, specifically: if two concepts appear in the same text, a link is generated between the two concepts, that is, two concepts appearing in the same text are connected with each other, that is, the concept nodes in the same text are all connected, step S4: and merging the initialization network and the user text network into a user concept network.
Specifically, after an initialization network and a user text network are respectively constructed, the nodes and links in the user text network are supplemented on the basis of the initialization network to form a complete user concept network.
Step S5: and constructing a text concept network corresponding to each text to be recommended according to the matching condition of the text to be recommended in the text library to be recommended and the concept library, wherein the nodes of the text concept network corresponding to the text to be recommended are used for representing the concepts matched with the concept library in the text to be recommended, and the edges of the text concept network corresponding to the text to be recommended represent the links between the concepts matched with the concept library.
Step S6: constructing a total concept network according to the text concept network corresponding to each text to be recommended, wherein nodes in the total concept network are used for representing concepts in all the text concept networks, and edges in the total concept network are used for representing links among the concepts contained in the text concept networks;
specifically, after a corresponding text concept network is constructed for each text to be recommended, a "total concept network" needs to be constructed for all texts to be recommended in the text library to be recommended.
Wherein, the total concept network comprises concepts in all texts to be recommended, and the meaning of the side of the total concept network is as follows: and adding links of pairwise combination of concepts in each text to be recommended into the total concept network, and simultaneously calculating the occurrence frequency of each link as the weight of the link. The nodes in the total concept network are all concepts in the text concept network, and in addition, weights can be given to the nodes and the edges, the weight of the node represents the number of times that the concepts appear in the text library to be recommended, the edges (links) represent the co-occurrence (whether the concepts appear in the same text) between the concepts, and the weight of the link is the number of times that the link appears in the text library to be recommended.
It should be noted that, for the initialization network, the user text network, the user concept network, the text concept network and the general concept network referred to in the present invention, they may all be referred to as "text concept network" as a whole, and in the above network, if two concepts appear in the same text, a link is generated between the two concepts (the link between the concepts is also the co-occurrence of the concepts), that is, a plurality of concepts appearing in the same text at the same time are connected with each other, that is, concept nodes in the same text are all connected. A network in which a concept is a node and text co-occurs as a link is called a "text concept network". The construction process of the user text network has been described in detail in the foregoing steps, the method of constructing the initialization network according to the initial concept set and the text concept network according to the concept library is similar to the above steps, and will not be described in detail herein.
Step S7: and determining whether to recommend the text to be recommended according to the matching and linking conditions of concept nodes contained in the text concept network corresponding to the text to be recommended in the user concept network.
In one embodiment, step S7 includes:
if all concept nodes contained in the text concept network corresponding to the text to be recommended are matched with nodes in the user concept network, and all concept nodes contained in the text concept network corresponding to the text to be recommended are fully connected in the user concept network, the text to be recommended is not recommended, the text to be recommended is deleted from the text library to be recommended, a candidate text library is obtained, and otherwise, the text to be recommended is recommended.
Specifically, if all concept nodes contained in the text to be recommended appear in the concept network of the user and the nodes are fully connected in the concept network of the user, the text to be recommended is not recommended, and the text to be recommended is deleted from the text library to be recommended, so that a candidate text library is obtained: FRW.
In one embodiment, recommending a text to be recommended includes:
acquiring incompleteness of the text to be recommended and importance of missing content according to missing conditions of links and nodes in a text concept network corresponding to the text to be recommended;
and orderly recommending the texts to be recommended in the candidate text library according to the incompleteness of the texts to be recommended and the importance of the missing content.
Specifically, the incompleteness of the text concept network corresponding to the text to be recommended is analyzed, and the incompleteness of the incompletely connected text in the candidate text library is analyzed, wherein the incompleteness of the incompletely connected text is represented in two aspects, and one link with the incomplete missing connection is recorded as: Δ link, one node that is an incomplete loss of nodes is denoted as: a Δ node.
In one embodiment, obtaining the incompleteness of the text to be recommended and the importance of the missing content according to the missing condition of the link and the node in the text concept network corresponding to the text to be recommended includes:
determining the incompleteness of the text to be recommended according to the number of the missing nodes and the number of the missing links;
and determining the importance of the missing content according to the importance of the missing link and the importance of the missing node in the text concept network corresponding to the text to be recommended, wherein the importance of the missing link is determined by whether the two concept nodes connected by the missing link belong to the same category, and the importance of the missing node is determined according to the degree of the missing node in the total concept network.
Specifically, when recommending whether a text to be recommended is recommended, the following two aspects are considered:
first, a general concept network for judging the importance (D) of missing concepts Δnode Representing the degree of all missing nodes in the overall concept network, D Δnode The higher the value, the more important the node is
And secondly, constructing a user concept network and a text concept network constructed by each text to be recommended. When calculating the score of the text to be recommended, comparing the concepts and links contained in the text concept network corresponding to the text to be recommended with the user concept network, and judging the supplement effect of the text on the user concept network (the number N of missing nodes) Δnode And the number of missing links N Δlink )
In one embodiment, according to the incompleteness of the text to be recommended and the importance of the missing content, orderly recommending the text to be recommended in the candidate text library, which comprises the following steps:
quantifying the incompleteness of the text to be recommended and the importance of the missing content to respectively obtain corresponding calculation formulas, wherein the incompleteness calculation formula of the text to be recommended is as follows: s Δ ＝log 2 (N Δlink )+N Δnode The formula for calculating the importance of the missing content is I Δ ＝N T(link) +D Δnode ，N Δlink Indicates the number of missing links, N Δnode Indicates the number of missing nodes, N T(link) Denotes the sum of the criticalities of all missing links, D Δnode Representing the degree of all missing nodes in the overall conceptual network.
Will S Δ And I Δ Normalizing to obtain two numerical dimensions S of text recommendation Δ And I Δ And according to two numerical dimensions S Δ And I Δ Calculating a recommended text recommendation score S A Wherein, in the step (A),
according to S A The text in the candidate text library is recommended in order according to the numerical value of the text.
Specifically, the deletion penalty S is calculated separately for each article in the FRW (i.e., each text to be analyzed) Δ And absence of criticality I Δ . Deletion penalty S Δ The degree of incompleteness of the text is reflected, and the higher the degree of incompleteness, the more the recommendation tends not to be made. The number of missing links is represented by N Δlink Indicating that the number of missing nodes is N Δnode And (4) showing. Absence of essential I Δ The importance of the missing content of the text is reflected, and the more important the missing content is, the more recommendation is likely to be performed. Criticality is noted as 1 if the two concept nodes linked by the missing link Δ link belong to different categories of concepts, otherwise 0. The sum of the criticality of all missing links Δ link is denoted as N T(link) . The criticality of the missing node Δ node depends on its presence in the general concept network G W Degree of chaining (Degree) in (1), and the numerical value is marked as D Δnode The maximum value of the connectivity of any node of the general concept network is n-1, and n is G w The minimum value of the total number of nodes of (1) is 0.
The concepts of different categories are obtained by clustering all the concepts in the concept library constructed in step S1, and the specific implementation process is as follows:
N-X set including N vocabulary sets obtained by parsing corpus based on S1 1 ，X 2 ...X N }. For each vocabulary set (vocabulary set or concept) X in the vocabulary set N I ＝{x j ，x k .., counting each vocabulary x contained therein in turn j Frequency Fx of occurrence in other vocabulary sets j . Mix Fx j After taking logarithm, normalizing to obtain weight coefficient Tx j ：
Weighting factor Tx j X multiplied by pre-training j Word vector of
Get the correction vector, set X I The sum of the modified word vectors of each word in the set plus x i Word vector of itself
Resulting modified word vectors
When X is present I When the data is a null set, the data is transmitted to the mobile terminal,
and then clustering all the vocabularies according to the corrected word vector of each vocabulary. The clustering method uses hierarchical clustering to obtain a plurality of clusters.
Recommended score S of text A Is equal to S Δ And I Δ The distance from the text coordinate to the origin in a rectangular coordinate system of coordinates. After calculating the recommendation score of the recommended text, according to S A The text in the candidate text library FRW is ranked from high to low and then recommended in sequence.
Example two
Based on the same inventive concept, the present embodiment provides a content recommendation apparatus based on a text concept network, please refer to fig. 2, including:
a concept library construction module 201, configured to analyze each text included in a text set including m independent texts, and construct a concept library, where the concept library includes concepts, and m is a positive integer;
the initialization network construction module 202 is used for screening out public concepts according to the proportion of the concepts in the concept library appearing in m independent texts, analyzing the basic education teaching materials and extracting the basic concepts, combining the basic concepts and the public concepts into a user initial concept set, and constructing an initialization network according to the matching condition of the texts in the basic education teaching materials and the concepts in the user initial concept set, wherein the nodes of the initialization network are used for representing the concepts in the basic education teaching materials, which are matched with the user initial concept set, and the edges of the initialization network represent the links between the concepts matched with the user initial concept set;
the user text network building module 203 is used for acquiring a user text and building a user text network according to the matching condition of the user text and the concept library, wherein nodes in the user text network are used for representing concepts matched with the concept library in the user text, and edges in the user text network represent links between the concepts matched with the concept library;
a user concept network construction module 204, configured to merge the initialization network and the user text network into a user concept network;
the text concept network building module 205 is configured to build a text concept network corresponding to each text to be recommended according to a matching condition between the text to be recommended in the text library to be recommended and the concept library, where a node of the text concept network corresponding to the text to be recommended is used to represent a concept matched with the concept library in the text to be recommended, and an edge of the text concept network corresponding to the text to be recommended represents a link between the concept matched with the concept library;
a total concept network construction module 206, configured to construct a total concept network according to a text concept network corresponding to each text to be recommended, where nodes in the total concept network are used to represent concepts in all the text concept networks, and edges in the total concept network are used to represent links between the concepts;
and the recommending module 207 is configured to determine whether to recommend the text to be recommended according to a matching condition of the concept nodes included in the text concept network corresponding to the text to be recommended in the total concept network and a matching and linking condition of the concept nodes included in the text concept network corresponding to the text to be recommended in the user concept network.
In one embodiment, the concept library construction module is specifically configured to:
preprocessing each text contained in a text set containing m independent texts to obtain all sentences and vocabularies of a corpus formed by the m independent texts, wherein m is a positive integer;
taking each vocabulary of the corpus as a concept subject word, traversing all sentences and vocabularies, and bringing the vocabulary which commonly appears in the same sentence with the concept subject word into a vocabulary set corresponding to the concept subject word, wherein the vocabulary set comprises the concept subject word and vocabulary elements;
and screening vocabulary elements of each vocabulary set to construct a concept library.
In one embodiment, the concept library construction module is further configured to:
counting each vocabulary element x in the vocabulary set j And concept topic word x i The number z of the texts which appear together, wherein z is less than or equal to m;
and judging whether the text quantity z is greater than or equal to a first threshold value, if so, taking the vocabulary elements as effective vocabularies of the vocabulary set and keeping the vocabulary elements in the vocabulary set, otherwise, removing the vocabulary elements from the vocabulary set.
In one embodiment, the user text network construction module is further configured to:
judging whether the user text contains concepts in a concept library;
and taking the concepts contained in the user text as nodes, and taking a connecting line between every two concepts as an edge to construct a user text network.
In one embodiment, the recommendation module is specifically configured to:
if all concept nodes contained in the text concept network corresponding to the text to be recommended are matched with nodes in the user concept network, and all concept nodes contained in the text concept network corresponding to the text to be recommended are fully connected in the user concept network, the text to be recommended is not recommended, the text to be recommended is deleted from the text library to be recommended, a candidate text library is obtained, and otherwise, the text to be recommended is recommended.
In one embodiment, the recommendation module is further configured to:
acquiring incompleteness of the text to be recommended and importance of missing content according to missing conditions of links and nodes in a text concept network corresponding to the text to be recommended;
and orderly recommending the texts to be recommended in the candidate text library according to the incompleteness of the texts to be recommended and the importance of the missing content.
In one embodiment, the recommendation module is further configured to:
determining the incompleteness of the text to be recommended according to the number of the missing nodes and the number of the missing links;
determining the importance of the missing content according to the criticality of the missing link and the criticality of the missing node in the text concept network corresponding to the text to be recommended, wherein the criticality of the missing link is determined by whether two concept nodes connected by the missing link belong to the same category, and the criticality of the missing node is determined according to the degree of the missing node in the total concept network.
In one embodiment, the recommendation module is further configured to:
quantifying the incompleteness of the text to be recommended and the importance of the missing content to respectively obtain corresponding calculation formulas, wherein the incompleteness calculation formula of the text to be recommended is as follows: s Δ ＝log 2 (N Δlink )+N Δnode The formula for calculating the importance of the missing content is, Δ ＝N T(link) +D Δnode ，N Δlink indicates the number of missing links, N Δnode Indicates the number of missing nodes, N T(link) Denotes the sum of the criticalities of all missing links, D Δnode Representing the degree of all missing nodes in the overall concept network;
will S Δ And the combination of (a) and (b), Δ normalizing to obtain two numerical dimensions S recommended by texts Δ And the combination of (a) and (b), Δ and according to two numerical dimensions S Δ And (c) and (d), Δ calculating a recommended text recommendation score S A Wherein, in the process,
according to S A The text in the candidate text library is recommended in order according to the numerical value of the text.
EXAMPLE III
Referring to fig. 3, based on the same inventive concept, the present application further provides a computer-readable storage medium 300, on which a computer program 311 is stored, which when executed implements the method according to the first embodiment.
Since the computer-readable storage medium introduced in the third embodiment of the present invention is a computer device used for implementing the content recommendation method based on the text concept network in the first embodiment of the present invention, based on the method introduced in the first embodiment of the present invention, those skilled in the art can understand the specific structure and modification of the computer-readable storage medium, and therefore, no further description is given here. Any computer readable storage medium used in the method of the first embodiment of the present invention is within the scope of the present invention.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the invention.
It will be apparent to those skilled in the art that various modifications and variations can be made in the embodiments of the present invention without departing from the spirit or scope of the embodiments of the invention. Thus, if such modifications and variations of the embodiments of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to encompass such modifications and variations.
Claims (6)
1. A content recommendation method based on a text concept network is characterized by comprising the following steps:
analyzing each text contained in a text set containing m independent texts, and constructing a concept library, wherein the concept library comprises concepts, and m is a positive integer;
screening out public concepts according to the proportion of concepts in the concept library appearing in m independent texts, analyzing a basic education teaching material, extracting basic concepts, combining the basic concepts and the public concepts into a user initial concept set, and constructing an initialization network according to the matching condition of texts in the basic education teaching material and the concepts in the user initial concept set, wherein nodes of the initialization network are used for representing the concepts in the basic education teaching material matched with the user initial concept set, and edges of the initialization network represent links between the concepts matched with the user initial concept set;
acquiring a user text, and constructing a user text network according to the matching condition of the user text and a concept library, wherein nodes in the user text network are used for representing concepts matched with the concept library in the user text, and edges in the user text network represent links between the concepts matched with the concept library;
merging the initialization network and the user text network into a user concept network;
constructing a text concept network corresponding to each text to be recommended according to the matching condition of the text to be recommended in the text library to be recommended and the concept library, wherein nodes of the text concept network corresponding to the text to be recommended are used for representing concepts matched with the concept library in the text to be recommended, and edges of the text concept network corresponding to the text to be recommended represent links between the concepts matched with the concept library;
constructing a total concept network according to the text concept network corresponding to each text to be recommended, wherein nodes in the total concept network are used for representing concepts in all the text concept networks, and edges in the total concept network are used for representing links among the concepts contained in the text concept networks;
determining whether to recommend the text according to the matching condition of concept nodes contained in the text concept network corresponding to the text to be recommended in the total concept network and the matching and linking condition of the concept nodes contained in the text concept network corresponding to the text to be recommended in the user concept network;
the method for determining whether to recommend the text comprises the following steps of determining whether to recommend the text according to the matching condition of concept nodes contained in a text concept network corresponding to the text to be recommended in a total concept network and the matching and linking condition of the concept nodes contained in the text concept network corresponding to the text to be recommended in a user concept network, wherein the method comprises the following steps:
if all concept nodes contained in the text concept network corresponding to the text to be recommended are matched with nodes in the user concept network and all the concept nodes contained in the text concept network corresponding to the text to be recommended are fully connected in the user concept network, not recommending the text to be recommended, deleting the text to be recommended from the text library to be recommended to obtain a candidate text library, and otherwise, recommending the text to be recommended;
recommending the text to be recommended, wherein the recommending comprises the following steps:
acquiring incompleteness of the text to be recommended and importance of missing content according to missing conditions of links and nodes in a text concept network corresponding to the text to be recommended;
sequentially recommending the texts to be recommended in the candidate text library according to the incompleteness of the texts to be recommended and the importance of the missing content;
according to the missing conditions of links and nodes in a text concept network corresponding to the text to be recommended, the incompleteness of the text to be recommended and the importance of the missing content are obtained, and the method comprises the following steps:
determining the incompleteness of the text to be recommended according to the number of the missing nodes and the number of the missing links;
determining the importance of the missing content according to the importance of the missing link and the importance of the missing node in the text concept network corresponding to the text to be recommended, wherein the importance of the missing link is determined by whether two concept nodes connected by the missing link belong to the same category, and the importance of the missing node is determined according to the degree of the missing node in the total concept network;
according to the incompleteness of the text to be recommended and the importance of the missing content, orderly recommending the text to be recommended in the candidate text library, wherein the orderly recommending comprises the following steps:
quantifying the incompleteness of the text to be recommended and the importance of the missing content to respectively obtain corresponding calculation formulas, wherein the incompleteness calculation formula of the text to be recommended is as follows: s Δ ＝log 2 (N Δlink )+N Δnode The formula for calculating the importance of the missing content is I Δ ＝N T(link) +D Δnode ，N Δlink Indicates the number of missing links, N Δnode Indicates the number of missing nodes, N T(link) Denotes the sum of the criticalities of all missing links, D Δnode Representing the degree of all missing nodes in the overall concept network;
will S Δ And I Δ Normalizing to obtain two numerical dimensions S of text recommendation Δ And I Δ And according to two numerical dimensions S Δ And I Δ Calculating a recommended text recommendation score S A Wherein, in the step (A),
according to S A The text in the candidate text library is recommended in order according to the numerical value of the text.
2. The method of claim 1, wherein parsing each text included in a collection of texts including m independent texts to construct a concept library comprises:
preprocessing each text contained in a text set containing m independent texts to obtain all sentences and vocabularies of a corpus formed by the m independent texts, wherein m is a positive integer;
taking each vocabulary of the corpus as a concept subject word, traversing all sentences and vocabularies, and bringing the vocabulary which commonly appears in the same sentence with the concept subject word into a vocabulary set corresponding to the concept subject word, wherein the vocabulary set comprises the concept subject word and vocabulary elements;
and screening vocabulary elements of each vocabulary set to construct a concept library.
3. The method of claim 2, wherein performing vocabulary element screening on each vocabulary set to build a concept base comprises:
counting each vocabulary element x in the vocabulary set j And concept topic word x i The number z of the texts which appear together, wherein z is less than or equal to m;
and judging whether the text quantity z is larger than or equal to a first threshold value, if so, taking the vocabulary elements as effective vocabularies of the vocabulary set and keeping the effective vocabularies in the vocabulary set, otherwise, removing the vocabulary elements from the vocabulary set.
4. The method of claim 1, wherein constructing a user text network based on matching user text to a concept library comprises:
judging whether the user text contains concepts in a concept library;
and taking concepts contained in the user text as nodes, and taking a connecting line between every two concepts as an edge to construct a user text network.
5. A content recommendation apparatus based on a text concept network, comprising:
the concept library construction module is used for analyzing each text contained in a text set containing m independent texts and constructing a concept library, wherein the concept library comprises concepts, and m is a positive integer;
the system comprises an initialization network construction module, a basic education teaching material and an initialization network, wherein the initialization network construction module is used for screening out public concepts according to the proportion of concepts in a concept library appearing in m independent texts, analyzing the basic education teaching material and extracting basic concepts, merging the basic concepts and the public concepts into a user initial concept set, and constructing the initialization network according to the matching condition of texts in the basic education teaching material and the concepts in the user initial concept set, wherein nodes of the initialization network are used for representing the concepts in the basic education teaching material, which are matched with the user initial concept set, and edges of the initialization network represent links between the concepts matched with the user initial concept set;
the user text network construction module is used for acquiring a user text and constructing a user text network according to the matching condition of the user text and the concept library, wherein nodes in the user text network are used for representing concepts matched with the concept library in the user text, and edges in the user text network represent links between the concepts matched with the concept library;
the user concept network construction module is used for combining the initialization network and the user text network into a user concept network;
the text concept network construction module is used for constructing a text concept network corresponding to each text to be recommended according to the matching condition of the text to be recommended in the text library to be recommended and the concept library, wherein the nodes of the text concept network corresponding to the text to be recommended are used for representing the concepts matched with the concept library in the text to be recommended, and the edges of the text concept network corresponding to the text to be recommended represent the links between the concepts matched with the concept library;
the general concept network construction module is used for constructing a general concept network according to the text concept network corresponding to each text to be recommended, nodes in the general concept network are used for representing concepts in all the text concept networks, and edges in the general concept network are used for representing links among the concepts;
the recommendation module is used for determining whether to recommend the text to be recommended according to the matching condition of the concept nodes contained in the text concept network corresponding to the text to be recommended in the total concept network and the matching and linking condition of the concept nodes contained in the text concept network corresponding to the text to be recommended in the user concept network;
wherein, the recommending module is further used for:
if all concept nodes contained in the text concept network corresponding to the text to be recommended are matched with nodes in the user concept network and all the concept nodes contained in the text concept network corresponding to the text to be recommended are fully connected in the user concept network, not recommending the text to be recommended, deleting the text to be recommended from the text library to be recommended to obtain a candidate text library, and otherwise, recommending the text to be recommended;
recommending the text to be recommended, wherein the recommending comprises the following steps:
acquiring incompleteness of the text to be recommended and importance of missing content according to missing conditions of links and nodes in a text concept network corresponding to the text to be recommended;
sequentially recommending the texts to be recommended in the candidate text library according to the incompleteness of the texts to be recommended and the importance of the missing contents;
according to the missing conditions of links and nodes in a text concept network corresponding to the text to be recommended, the incompleteness of the text to be recommended and the importance of the missing content are obtained, and the method comprises the following steps:
determining the incompleteness of the text to be recommended according to the number of the missing nodes and the number of the missing links;
determining the importance of the missing content according to the criticality of the missing link and the criticality of the missing node in the text concept network corresponding to the text to be recommended, wherein the criticality of the missing link is determined by whether two concept nodes connected by the missing link belong to the same category, and the criticality of the missing node is determined according to the degree of the missing node in the total concept network;
according to the incompleteness of the text to be recommended and the importance of the missing content, orderly recommending the text to be recommended in the candidate text library, wherein the orderly recommending comprises the following steps:
quantifying the incompleteness of the text to be recommended and the importance of the missing content to respectively obtain corresponding calculation formulas, wherein the incompleteness calculation formula of the text to be recommended is as follows: s. the Δ ＝log 2 (N Δlink )+N Δnode The formula for calculating the importance of the missing content is I Δ ＝N T(link) +D Δnode ，N Δlink Indicates the number of missing links, N Δnode Indicates the number of missing nodes, N T(link) Denotes the sum of the criticalities of all missing links, D Δnode Representing the degree of all missing nodes in the overall concept network;
will S Δ And I Δ Normalizing to obtain two numerical dimensions S recommended by texts Δ And I Δ And according to two numerical dimensions S Δ And I Δ Calculating a recommended text recommendation score S A Wherein, in the step (A),
according to S A The text in the candidate text library is recommended in order according to the numerical value of the text.
6. A computer-readable storage medium, on which a computer program is stored, characterized in that the program, when executed, implements the method of any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911258023.1A CN112948568B (en) | 2019-12-10 | 2019-12-10 | Content recommendation method and device based on text concept network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911258023.1A CN112948568B (en) | 2019-12-10 | 2019-12-10 | Content recommendation method and device based on text concept network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112948568A CN112948568A (en) | 2021-06-11 |
| CN112948568B true CN112948568B (en) | 2022-08-30 |
Family
ID=76225475
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911258023.1A Active CN112948568B (en) | 2019-12-10 | 2019-12-10 | Content recommendation method and device based on text concept network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112948568B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104765730A (en) * | 2014-01-02 | 2015-07-08 | 株式会社理光 | Method and device for recommending interested people |
| CN110020189A (en) * | 2018-06-29 | 2019-07-16 | 武汉掌游科技有限公司 | A kind of article recommended method based on Chinese Similarity measures |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080294622A1 (en) * | 2007-05-25 | 2008-11-27 | Issar Amit Kanigsberg | Ontology based recommendation systems and methods |
| US20100281025A1 (en) * | 2009-05-04 | 2010-11-04 | Motorola, Inc. | Method and system for recommendation of content items |
| US20120123992A1 (en) * | 2010-11-11 | 2012-05-17 | Rovi Technologies Corporation | System and method for generating multimedia recommendations by using artificial intelligence concept matching and latent semantic analysis |
| US10437869B2 (en) * | 2014-07-14 | 2019-10-08 | International Business Machines Corporation | Automatic new concept definition |
| US10592807B2 (en) * | 2016-09-01 | 2020-03-17 | Facebook, Inc. | Systems and methods for recommending content items |
-
2019
- 2019-12-10 CN CN201911258023.1A patent/CN112948568B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104765730A (en) * | 2014-01-02 | 2015-07-08 | 株式会社理光 | Method and device for recommending interested people |
| CN110020189A (en) * | 2018-06-29 | 2019-07-16 | 武汉掌游科技有限公司 | A kind of article recommended method based on Chinese Similarity measures |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112948568A (en) | 2021-06-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111563384B (en) | Evaluation object identification method and device for E-commerce products and storage medium | |
| CN112070138B (en) | Construction method of multi-label mixed classification model, news classification method and system | |
| CN104978314B (en) | Media content recommendations method and device | |
| CN111104526A (en) | Financial label extraction method and system based on keyword semantics | |
| CN112241626A (en) | Semantic matching and semantic similarity model training method and device | |
| CN109325146A (en) | A kind of video recommendation method, device, storage medium and server | |
| CN112199602B (en) | Post recommendation method, recommendation platform and server | |
| CN114254201A (en) | Recommendation method for science and technology project review experts | |
| CN111966832A (en) | Evaluation object extraction method and device and electronic equipment | |
| CN112182145A (en) | Text similarity determination method, device, equipment and storage medium | |
| CN111782793A (en) | Intelligent customer service processing method, system and equipment | |
| CN112015907A (en) | Method and device for quickly constructing discipline knowledge graph and storage medium | |
| CN112214984A (en) | Content plagiarism identification method, device, equipment and storage medium | |
| CN110287493B (en) | Risk phrase identification method and device, electronic equipment and storage medium | |
| CN108763221B (en) | Attribute name representation method and device | |
| CN110019820A (en) | Main suit and present illness history symptom Timing Coincidence Detection method in a kind of case history | |
| Chenze et al. | Iterative approach for novel entity recognition of foods in social media messages | |
| CN111259223B (en) | News recommendation and text classification method based on emotion analysis model | |
| CN112948568B (en) | Content recommendation method and device based on text concept network | |
| CN111178080A (en) | Named entity identification method and system based on structured information | |
| CN115859963A (en) | Similarity judgment method and system for new word and semantic recommendation | |
| CN112257439B (en) | Method and device for mining hot root words through public opinion data | |
| CN113688633A (en) | Outline determination method and device | |
| Lubis et al. | Improving course review helpfulness Prediction through sentiment analysis | |
| CN113988057A (en) | Title generation method, device, equipment and medium based on concept extraction |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |