WO2021098648A1 - Procédé, appareil et dispositif de recommandation de texte, et support - Google Patents

Procédé, appareil et dispositif de recommandation de texte, et support Download PDF

Info

Publication number
WO2021098648A1
WO2021098648A1 PCT/CN2020/129115 CN2020129115W WO2021098648A1 WO 2021098648 A1 WO2021098648 A1 WO 2021098648A1 CN 2020129115 W CN2020129115 W CN 2020129115W WO 2021098648 A1 WO2021098648 A1 WO 2021098648A1
Authority
WO
WIPO (PCT)
Prior art keywords
text
preset
candidate
node
target user
Prior art date
Application number
PCT/CN2020/129115
Other languages
English (en)
Chinese (zh)
Inventor
蔡远航
郑少杰
付勇
范增虎
Original Assignee
深圳前海微众银行股份有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 深圳前海微众银行股份有限公司 filed Critical 深圳前海微众银行股份有限公司
Publication of WO2021098648A1 publication Critical patent/WO2021098648A1/fr

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
    • G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
    • G06F16/367—Ontology
    • 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

Definitions

  • This application relates to the technical field of financial technology (Fintech), in particular to a text recommendation method, device, equipment and medium.
  • the main purpose of this application is to provide a text recommendation method, device, device, and medium, which aims to solve the technical problems of excessive recommendation and low recommendation accuracy in the current process of content recommendation based on keywords.
  • an embodiment of the present application provides a text recommendation method, and the text recommendation method includes:
  • More than one updated text containing at least one of the keywords is retrieved from a preset text database set as the first candidate text;
  • the affair map includes the association relationship between the text and the text, and each association relationship has its corresponding degree of association;
  • the selected text is filtered out from the first candidate text and the second candidate text, and the selected text is recommended to the target user.
  • the recalling a preset affair atlas, and selecting from the preset text database set according to the preset affair atlas, the total relevance to the first candidate text is not less than a preset Before the update text step of the associated threshold includes:
  • the preset affair atlas is generated according to the preprocessed text.
  • the step of generating the preset affair atlas according to the preprocessed text includes:
  • a plurality of preset text association relationships are recognized for each clause in the clause list to obtain the node text to be processed, where the preset text association relationships include, but are not limited to, succession, cause and effect, condition, and parallelism relationship;
  • Two to-be-processed node texts whose node distance is less than the first preset distance are subjected to iterative grafting processing until each of the to-be-processed node texts is in a convergent state where the node text relationship edge no longer changes, wherein The text of each node to be processed in the state is set as the text of the convergent node;
  • the preset affair graph is generated.
  • the update text of the associated threshold, as the second candidate text includes:
  • the second node distance is selected from the third candidate text to be less than the second preset distance, and the convergent node text corresponding to the second preset distance is the first target text of the node text that the user pays attention to, or from the first target text
  • the second node distance selected from the three candidate texts is less than the second preset distance, and the second target text of the user-focused node text exists in the preset filtering logic depth range of the convergent node text corresponding to the second preset distance, where ,
  • the depth of the screening logic is determined according to the degree of relevance of the respective relevance relationships in the affair map;
  • the step of filtering out the selected text from the first candidate text and the second candidate text according to the operation behavior, and recommending the selected text to the target user includes :
  • the selected text is selected from the first candidate text and the second candidate text, and the selected text is recommended to the target user .
  • the step of obtaining the relevance of each text to the target user includes:
  • the word position includes the first sentence position of the first paragraph of the text, the first sentence position of the end paragraph of the text, the position of the first sentence of the text, the position of the first sentence of the end of the text, the position of the first sentence of the non-first paragraph, and the first sentence of the non-final paragraph. position;
  • the step of obtaining the preference degree of the target user according to the operation behavior includes:
  • the first Pearson correlation coefficient between the second document vector and the first document vector is acquired, and the preference of the target user is acquired according to the first Pearson correlation coefficient.
  • the step of obtaining the first document vector of each text in the historical browsing text includes:
  • the step of obtaining a first Pearson correlation coefficient between the second document vector and the first document vector, and obtaining the preference degree of the target user according to the first Pearson correlation coefficient include:
  • the selected text is selected from the first candidate text and the second candidate text according to the amount of communication, the degree of relevance, and the degree of preference, and the selected text is selected from among the first candidate text and the second candidate text.
  • the steps of selecting text to recommend to the target user include:
  • a preset number of texts are selected in order from high to low as the selected text, and the selected text is recommended to the target user.
  • the present application also provides a text recommendation device, the text recommendation device includes:
  • the monitoring module is used to monitor the operation behavior of the target user, and determine the keywords associated with the target user according to the operation behavior;
  • a retrieval module configured to retrieve more than one updated text containing at least one of the keywords from a preset text database set as the first candidate text;
  • the selection module is configured to retrieve a preset affair atlas, and select from the preset text database set according to the preset affair atlas with a total relevance to the first candidate text that is not less than a preset relevance threshold Update the text, as the second candidate text, the affair map includes the association relationship between the text and the text, and each association relationship has its corresponding degree of association;
  • the screening module is configured to screen out the selected text from the first candidate text and the second candidate text according to the operation behavior, and recommend the selected text to the target user.
  • the text recommendation device further includes:
  • the collection module is configured to collect the text to be processed from the preset text database collection every preset time period;
  • the preprocessing module is used to perform html tag filtering, symbol filtering, and sentence processing on the to-be-processed text through a preset regular expression to obtain a preprocessed text composed of a sentence list;
  • the generating module is used to generate the preset affair atlas according to the preprocessed text.
  • the generating module includes:
  • the recognition unit is configured to recognize multiple preset text association relationships for each clause in the clause list to obtain the node text to be processed, where the preset text association relationships include, but are not limited to, Shun Cheng, Causality, conditions, and parallel relationships;
  • the first obtaining unit is configured to perform word segmentation processing on the node text to be processed by a preset word segmentation tool, and obtain the word vector of each word segmentation, and obtain the node vector of each node text to be processed based on the word vector of each word segmentation;
  • the first calculation unit is configured to calculate the first node distance between each to-be-processed node text and other to-be-processed node texts according to the node vector of each to-be-processed node text;
  • the grafting processing unit is configured to perform iterative grafting processing on two to-be-processed node texts whose node distance is less than a first preset distance, until each of the to-be-processed node texts is in a convergent state where the node text relationship edge no longer changes, wherein , Setting each of the node texts to be processed in a convergent state as the convergent node text;
  • the generating unit is configured to generate the preset affair graph based on the node text relationship edge between the convergent node text and the convergent node text.
  • the selection module includes:
  • the retrieval unit is configured to retrieve a preset affair map, and determine whether there is a convergent node text in the corresponding clause that contains the keyword in the affair map;
  • the first setting unit is configured to, if it exists, set the convergent node text containing the keyword in the corresponding clause as the user-focused node text, and update it from the preset text database within a preset time period Select a third candidate text other than the first candidate text from the text, and perform word segmentation processing on the title of each text in the third candidate text by a preset word segmentation tool to obtain a title vector of each text in the third candidate text;
  • a second calculation unit configured to calculate the second node distance between the title vector and the node vector of each convergent node text in the affair map
  • the selecting unit is configured to select, from the third candidate text, a second node distance less than a second preset distance, and the convergent node text corresponding to the second preset distance is the first target text of the node text that the user pays attention to; Or select the second node distance from the third candidate text to be less than the second preset distance, and the preset filtering logic depth range of the convergent node text corresponding to the second preset distance is less than the second node text of the user concerned.
  • Target text wherein the logical depth of the screening is determined according to the degree of relevance of the respective association relationships in the affair map;
  • the second setting unit is configured to set the first target text and the second target text as the second candidate text.
  • the screening module includes:
  • the second acquiring unit is configured to acquire the spread of each text in the first candidate text and the second candidate text, and acquire the relevance of each text to the target user, and acquire the information according to the operation behavior. State the preference of the target user;
  • a recommendation unit configured to filter out selected text from the first candidate text and the second candidate text according to the amount of communication, the degree of relevance, and the degree of preference, and recommend the selected text To the target user.
  • the second acquiring unit includes:
  • the first obtaining subunit is configured to obtain the number of times the keyword appears in each text of the first candidate text, and set the number of times as the number of words;
  • the second obtaining subunit is used to obtain the position where the keyword appears in each text of the first candidate text, set the position as a word position, and obtain a preset position weight corresponding to the word position , Where the word position is different, the position weight is different, the word position includes the first sentence position of the first paragraph of the text, the first sentence position of the last paragraph of the text, the position of the first sentence of the text, the position of the last sentence of the text, the position of the non-first paragraph Sentence position and the position of the first sentence of the non-final paragraph
  • the third acquisition subunit is used to acquire the ratio of the number of sentences between the first and last occurrences of the keyword in each text of the first candidate text to the total number of sentences in the full text, and compare all The ratio value is set as the word span;
  • the fourth obtaining subunit is used to obtain the target text between the first and last occurrences of the keyword in each text of the first candidate text, and obtain the average preset value in the target text.
  • the number of sentences contains the number of the keywords, and the average number of the keywords contained in each preset number of sentences is set as the word density;
  • the fifth acquiring subunit is configured to acquire the first degree of relevance of each text in the first candidate text according to the number of words, the preset position weight corresponding to the word position, the word span and the word density ;
  • the sixth obtaining subunit is configured to obtain the screening logic depth of each text in the second candidate text, and determine the second relevance of each text in the second candidate text according to the screening logic depth.
  • the second acquiring unit includes:
  • the seventh obtaining subunit is configured to obtain the historical browsing text of the target user from the operation behavior, obtain the first document vector of each text in the historical browsing text, and obtain the first candidate text and all the texts. State the second document vector of each text in the second candidate text;
  • the eighth obtaining subunit is configured to obtain the first Pearson correlation coefficient between the second document vector and the first document vector, and obtain the preference degree of the target user according to the first Pearson correlation coefficient.
  • the seventh acquiring subunit is used to implement:
  • the eighth acquiring subunit is used to implement:
  • the screening module includes:
  • the third calculation unit is configured to calculate each of the first candidate text and the second candidate text according to the amount of communication, the first degree of relevance, the second degree of relevance, and the preference degree Value score
  • the screening unit is configured to select a preset number of texts as selected texts in order from high to low according to the value score, and recommend the selected texts to the target user.
  • the application also provides a medium on which a text recommendation program is stored, and when the text recommendation program is executed by a processor, the steps of the text recommendation method described above are implemented.
  • This application monitors the operation behavior of the target user, and determines the keywords associated with the target user according to the operation behavior; after obtaining the keywords, retrieves more than one that contains at least one of the keywords from the preset text database collection The updated text of, as the first candidate text; after the first candidate text is obtained, the preset affair atlas is retrieved, and the preset affair atlas is selected from the preset text database collection and the first The updated text whose total relevance of the candidate text is not less than the preset relevance threshold is used as the second candidate text.
  • the acquisition of the second candidate text expands the selection category of the candidate text in the recommendation process.
  • the affair map includes the text and the text Each association relationship has a corresponding degree of association; according to the operation behavior, the selected text is selected from the first candidate text and the second candidate text, and the selected text is recommended to The target user. That is, in this application, the selected text is not only selected from the first candidate text searched based on keywords, but from the second candidate text and the first candidate text obtained from the preset affair graph, etc. The selected text is selected from the collection, thus avoiding the simplification of content recommendation, and because the association between the reference text and the text in this application is used to recommend content instead of just recommending based on keywords, this application can Improve recommendation accuracy.
  • Fig. 1 is a schematic flowchart of a first embodiment of a text recommendation method for this application
  • Figure 2 is a second embodiment of the text recommendation method of this application based on recalling a preset affair map, and selecting a total of the first candidate text from the preset text database set according to the preset affair map.
  • FIG. 3 is a schematic diagram of the device structure of the hardware operating environment involved in the method of the embodiment of the present application.
  • Figure 4 is a schematic diagram of the first scenario involved in the text recommendation method of this application.
  • Figure 5 is a schematic diagram of a second scenario involved in the text recommendation method of this application.
  • FIG. 6 is a schematic diagram of a third scenario involved in the text recommendation method of this application.
  • FIG. 7 is a schematic diagram of a fourth scenario involved in the text recommendation method of this application.
  • FIG. 8 is a schematic diagram of a fifth scenario involved in the text recommendation method of this application.
  • FIG. 9 is a schematic diagram of a sixth scenario involved in the text recommendation method of this application.
  • FIG. 10 is a schematic diagram of a seventh scenario involved in the text recommendation method of this application.
  • FIG. 11 is a schematic diagram of an eighth scene involved in the text recommendation method of this application.
  • FIG. 12 is a schematic diagram of a ninth scene involved in the text recommendation method of this application.
  • FIG. 13 is a schematic diagram of the process involved in the text recommendation method of the present application.
  • the text recommendation method includes:
  • Step S10 monitoring the operation behavior of the target user, and determining keywords associated with the target user according to the operation behavior;
  • Step S20 retrieve more than one updated text containing at least one of the keywords from a preset text database set as the first candidate text;
  • Step S30 retrieve a preset affair atlas, and select updated texts whose total relevance to the first candidate text is not less than a preset relevance threshold from the preset text database set according to the preset affair atlas ,
  • the affair map includes the association relationship between the text and the text, and each association relationship has its corresponding degree of association;
  • Step S40 According to the operation behavior, the selected text is selected from the first candidate text and the second candidate text, and the selected text is recommended to the target user.
  • Step S10 monitoring the operation behavior of the target user, and determining keywords associated with the target user according to the operation behavior;
  • the current public opinion system generally achieves the above functions through the following process: 1.
  • Data collection Collect all information sources on the entire network, including news media, forums, blogs, and microblogs. And various information clients, etc.; 2.
  • Data filtering filter news data according to the keywords of the monitoring task configured on the public opinion system, for example, if the text of a news article contains keywords configured by the user, the article will be retained News is used for subsequent processing; 3.
  • Data processing For all news containing keywords, sequentially calculate the emotional tendency of the text, the amount of news dissemination, and the degree of correlation between the news text and the keywords, etc.; 4. Data push: comprehensively consider the news The sentiment, the amount of communication, the degree of relevance, and the user’s click preference on the historical push news data on the public opinion system, the news data processed in steps 1-3 are sorted, and the multiple news that users are most likely to be interested in are selected for push. That is, in the prior art, the user’s configured keywords and the user’s click preferences for historical news push data are completely relied upon to recommend target content.
  • word2vec one This kind of word vector model can describe the semantic similarity of Chinese vocabulary according to the distance between word vectors.
  • word2vec word vectors Preference is narrowly defined, which leads to the simplification of the content of the push. Specifically, for example, a user clicks on a piece of news such as "Enterprise A and Bank B have reached a cooperation" several times in a certain period of time. After processing word2vec word vector, the public opinion system learns that the user prefers Enterprise A and Bank B. News from time to time.
  • the operation behavior of the target user is monitored, and the keyword associated with the target user is determined according to the operation behavior.
  • the operation behavior includes a sliding behavior or a search behavior triggered by input of a keyword. If the operation behavior is It is a sliding behavior, and determining the keywords associated with the target user according to the operation behavior may be: extracting pre-stored keywords associated with the target user, and the pre-stored keywords associated with the target user are obtained based on the historical browsing text of the target user If the operation behavior is a search behavior of inputting a keyword, the keyword associated with the target user may be the input keyword, or may be a combination of the input keyword and a pre-stored keyword associated with the target user.
  • Step S20 retrieve more than one updated text containing at least one of the keywords from a preset text database set as the first candidate text;
  • the preset text database collection includes a database collection composed of news media, forums, blogs, microblogs, and other various information clients.
  • More than one updated text containing at least one of the keywords is retrieved from a preset text database set as the first candidate text, and more than one that contains at least one of the keywords is retrieved from a preset text database set
  • the updated text of, as the first candidate text may be: searching for more than one updated text containing at least one of the keywords from a preset text database set in real time as the first candidate text (for real-time recommendation), or every More than one updated text containing at least one of the keywords is retrieved from the preset text database set in real time at intervals of a certain period of time, as the first candidate text (for timing recommendation), or just from the preset text this time More than one updated text containing at least one of the keywords is retrieved from the database set as the first candidate text (for the target user to make recommendations when searching), etc.
  • Step S30 retrieve a preset affair atlas, and select updated texts whose total relevance to the first candidate text is not less than a preset relevance threshold from the preset text database set according to the preset affair atlas ,
  • the affair map includes the association relationship between the text and the text, and each association relationship has its corresponding degree of association;
  • the preset affair map has been generated and updated in real time or regularly.
  • the total relevance can be determined by the logical depth between words or the distance between words, for example, the total relevance is not less than the preset relevance threshold. It can be that the logical depth is not less than 2 depth units or the distance between words is not less than 10 preset distance units, etc., and the relationship between text and text can be causality, continuity, etc., causality, continuation, etc. The degree of association is different.
  • the second candidate text improves the recommendation range in the content recommendation process.
  • the preset affair atlas is retrieved, and the total relevance to the first candidate text is selected from the preset text database set according to the preset affair atlas.
  • the steps include:
  • Step A1 Collect the text to be processed from the preset text database collection every preset time period;
  • Step A2 performing html tag filtering, symbol filtering, and sentence processing on the to-be-processed text through a preset regular expression to obtain a preprocessed text composed of a sentence list;
  • Step A3 Generate the preset affair atlas according to the preprocessed text.
  • the text to be processed is collected from the preset text database collection every preset time period (which may include real-time). Among them, since the volume of text to be processed collected every day is on the order of tens of millions, it can be used
  • the preset collection model such as the preset spark streaming model, completes the collection.
  • html tag filtering, symbol filtering, and sentence processing are performed on the text to be processed using preset regular expressions to obtain the preprocessed text composed of a list of sentences. In one embodiment, each interval is pre-processed.
  • the step of generating the preset affair atlas according to the preprocessed text includes:
  • Step A31 Recognize multiple preset text association relationships for each clause in the clause list to obtain the node text to be processed, where the preset text association relationships include but are not limited to succession, cause and effect, Conditions and juxtaposition;
  • a plurality of preset text association relations are recognized, and the preset text association relations include but are not limited to types such as succession, causality, condition, and parallel relations.
  • the node text to be processed is obtained.
  • two events that express a succession/causality/condition/parallel relationship can be identified from each text sentence of each text of the preprocessed text, and two events
  • the phrase can be set as the text of the node to be processed in the affair map, and the affair map uses a directed edge (pointing line) to connect the two node texts to be processed, such as "The central bank's interest rate cut will make the loan cost lower"
  • the node text to be processed as shown in Figure 4 can be obtained from the text sentence of the preprocessed text.
  • the model can also be extracted based on the preset text association relationship, and the table sequence can be identified from each text sentence of each text of the preprocessed text.
  • the preset word combinations that express succession in the preset text association relationship are: (first, second), (first, then), (on one hand, on the other hand ), (first, then), (first, then), (first, then), etc., if a sentence contains both the above-mentioned words that express the succession and two words in a certain phrase are combined, and the two words are in the sentence The order of appearance is consistent with the order defined in the phrase.
  • the two clauses guided by these two words are extracted through the preset guiding clause model in the pre-set text association extraction model, and all punctuation marks in the clauses are removed ( Preset), modal particles (preset), auxiliary words (preset) and stop words (preset), etc., as the two to-be-processed node texts in the affair map, and use a table to inherit the Toward the edge (the edge of the logical relationship) connects the text of the two nodes to be processed. For example, the sentence "A first, then B" will be shown in Figure 5 after processing.
  • the pre-set word combinations that express cause and effect are: (because, so), (because, cause), (because, cause), (because, therefore), (because, so), (because, cause) , (Because, make), (because, therefore), (since, then), (since, then), (once, then), (because, therefore), (because, so), (because, cause) , (Because, therefore), (because, cause), (because, therefore), (_, therefore), (_, cause), (_, thus), (_, cause), ( _, therefore) etc.
  • the underscore "_" in the phrase indicates empty words, and the matching of empty words can be ignored in the subsequent matching process.
  • a sentence contains both the above causal words and two words in a phrase, and the order of appearance of the two words in the sentence is consistent with the order defined in the phrase, the two words are combined through the preset guiding clause model.
  • Two clauses guided by a word are extracted, and all punctuation marks (predetermined), modal particles (predetermined), auxiliary words (predetermined) and stop words (predetermined) in the clauses are removed, etc.
  • a directed edge edge of logical relationship
  • the preset word combinations for expressing conditions are: (if, then), (if, then), (if, then), (if, then), (if, then), (if, just), ( If, then), (if, then), (once, then), (as long as, then), (if, then), (only, only), etc. If a sentence contains the words of the above table conditions and combines two words in a certain phrase, and the order of appearance of the two words in the sentence is consistent with the order defined in the phrase, the two words are combined through the preset guiding clause model.
  • the preset combinations of words in the table are: (not only, but also), (not only, and), (not only, but also), (not only, also), (not only, but also), (not only, and), ( Not only, but also), (not only, but also), (not only, but also), (not only, and), (not only, but also), (not only, but also), (not only, but also), (not only, and), (not only, but also), (not only, but also), (not only, but also), (not only, but also), (not only, but also), (not only, and), (not only, Also), (not only, but also), (or, or), (or, or), (or, or), etc.
  • a sentence contains both words in the above table and combines two words in a phrase, and the order of appearance of the two words in the sentence is consistent with the order defined in the phrase, the two words are combined through the preset guiding clause model.
  • Two clauses guided by a word are extracted, and all punctuation marks (predetermined), modal particles (predetermined), auxiliary words (predetermined) and stop words (predetermined) in the clauses are removed, etc.
  • a directed edge the edge of the logical relationship
  • the preprocessed text includes more than a preset number of text data such as 100,000, you can use the preset bidirectional extraction network model to extract multiple preset text associations in each text sentence in the preprocessed text Relational event phrases.
  • Step A32 Perform word segmentation processing on the node text to be processed by a preset word segmentation tool, and obtain the word vector of each word segmentation, and obtain the node vector of each node text to be processed based on the word vector of each word segmentation;
  • a preset word segmentation tool such as a preset stuttering word segmentation tool (an open source Chinese word segmentation tool that can segment and mark the input Chinese text).
  • a preset word2vec a word vector model that maps each Chinese vocabulary to a high-dimensional vector (200-dimensional vector can be taken), assuming that there are 5 high-dimensional vectors, abcde, and these 5 word vectors according to the corresponding dimensions and dimensional weights By adding together, the word vector of the word segmentation can be obtained
  • the word vector of each word segmentation is obtained, and the node vector of each node text to be processed is obtained based on the word vector of each word segmentation.
  • Step A33 Calculate the first node distance between each to-be-processed node text and other to-be-processed node texts according to the node vector of each to-be-processed node text;
  • the first node distance between each to-be-processed node text and other to-be-processed node texts is calculated according to the node vector of each to-be-processed node text, specifically, the Pearson correlation of the preset node text is obtained
  • the coefficient calculation formula is to calculate the node text Pearson correlation coefficient between the two node text vectors to be processed according to the calculation formula of the node vector of each node text to be processed and the Pearson correlation coefficient of the preset node text.
  • ⁇ represents the Pearson correlation coefficient of the node text between two node text vectors to be processed, then the first node distance between the two node texts to be processed can be expressed as 1-( ⁇ +1)/2, for each node to be processed Process the node text, and calculate the distance between the text of the node to be processed and all other texts of the node to be processed in turn.
  • Step A34 Perform iterative grafting processing on the two to-be-processed node texts whose node distance is less than the first preset distance, until each of the to-be-processed node texts is in a convergent state where the node text relationship edge no longer changes, where all The text of each node to be processed in the convergent state is set as the text of the convergent node;
  • Step A35 Generate the preset affair graph based on the node text relationship edge between the convergent node text and the convergent node text.
  • Two to-be-processed node texts whose node distance is less than the first preset distance are subjected to iterative grafting processing until each of the to-be-processed node texts is in a convergent state where the node text relationship edge no longer changes, wherein the convergent
  • the text of each node to be processed in the state is set as the convergent node text.
  • the node text to be processed All the relationships of A are grafted to the node text B to be processed, and the node text A to be processed is deleted at the same time, as shown in Figure 9, if the distance between the node text A to be processed and the node text C to be processed is less than the first preset distance as If it is less than 0.3, the relationship between the texts of the to-be-processed nodes as shown in Figure 10 is obtained, and the two texts of the to-be-processed nodes whose node distance is less than the first preset distance are grafted until each of the to-be-processed texts is to be processed.
  • the first preset distance such as less than 0.3
  • Processing node texts is in a convergent state, that is, iteratively executes the calculation process of grafting two to-be-processed node texts whose node distance is less than the first preset distance until each of the to-be-processed node texts of the to-be-processed text is at Convergence state, so that the relationship edge (relational boundary) of the affair map composed of the text of each node to be processed no longer changes, so a affair map with directed edges is generated, that is, the map is considered to have reached the state of convergence. In the process, it is also necessary to deal with the relational edges of the table "coupling" in the affair graph.
  • Step S40 According to the operation behavior, the selected text is selected from the first candidate text and the second candidate text, and the selected text is recommended to the target user.
  • This application monitors the operation behavior of the target user, and determines the keywords associated with the target user according to the operation behavior; after obtaining the keywords, retrieves more than one that contains at least one of the keywords from the preset text database collection The updated text of, as the first candidate text; after the first candidate text is obtained, the preset affair atlas is retrieved, and the preset affair atlas is selected from the preset text database collection and the first The updated text whose total relevance of the candidate text is not less than the preset relevance threshold is used as the second candidate text.
  • the acquisition of the second candidate text expands the selection category of the candidate text in the recommendation process.
  • the affair map includes the text and the text Each association relationship has a corresponding degree of association; according to the operation behavior, the selected text is selected from the first candidate text and the second candidate text, and the selected text is recommended to The target user. That is, in this application, the selected text is not only selected from the first candidate text searched based on keywords, but from the second candidate text and the first candidate text obtained from the preset affair graph, etc. The selected text is selected from the collection, thus avoiding the simplification of content recommendation, and because the association between the reference text and the text in this application is used to recommend content instead of just recommending based on keywords, this application can Improve recommendation accuracy.
  • this application provides another embodiment of the text recommendation method.
  • the preset affair atlas is retrieved, and the preset affair atlas is retrieved from the preset text database according to the preset affair atlas.
  • the step of selecting update texts whose total relevance to the first candidate text is not less than a preset relevance threshold from the set as the second candidate text includes:
  • Step S31 retrieve a preset affair map, and determine whether there is a convergent node text in the corresponding clause that contains the keyword in the affair map;
  • Step S32 if it exists, set the convergent node text containing the keyword in the corresponding clause as the user-focused node text, and select from the text updated within the preset time period from the preset text database
  • For the third candidate text other than the first candidate text perform word segmentation processing on the title of each text in the third candidate text by using a preset word segmentation tool to obtain the title vector of each text in the third candidate text;
  • the corresponding clause may or may not contain the key If there is no convergent node text containing the keyword in the corresponding clause in the affair map, no subsequent processing is performed, and the selected text can be directly selected from the first candidate text for recommendation.
  • the node text to be processed is marked as "user concerned node text", from the text updated within the preset time period from the preset text database
  • Select the third candidate text other than the first candidate text and perform word segmentation processing on the title of each text in the third candidate text by using a preset word segmentation tool to obtain the title vector of each text in the third candidate text.
  • the title of each text in the third candidate text is preset to stutter word segmentation, and the title vector of each text in the third candidate text is obtained with the help of the preset word2vec tool.
  • the purpose of obtaining the title vector is to calculate the distance of the second node.
  • Step S33 Calculate the second node distance between the title vector and the node vector of each convergent node text in the affair map;
  • Step S34 Select the second node distance from the third candidate text to be less than a second preset distance, and the convergent node text corresponding to the second preset distance is the first target text of the node text that the user pays attention to, or
  • the second node distance selected from the third candidate text is less than the second preset distance, and the second target of the user-focused node text exists in the preset filtering logic depth range of the convergent node text corresponding to the second preset distance Text, wherein the logical depth of the screening is determined according to the degree of relevance of the respective association relationships in the affair atlas;
  • the filtering logic depth is based on the relationship between the relationships in the affair map
  • the degree of relevance is determined, that is, the logical depth of screening can be defined as follows: the logical depth of the edge of the table "parallel" logical relationship is recorded as 0.5, the logical depth of the edge of the table “shuncheng” logical relationship is recorded as 0.7, and the logical depth of the table "causal” logical relationship and The logical depth of the edge of the table “condition” logical relationship is recorded as 1, and the logical depth between the texts of two nodes to be processed is the sum of the logical depths of all the edges between the nodes, for example, the text of the B to be processed node
  • Step S35 Set the first target text and the second target text as the second candidate text.
  • the first target text and the second target text are obtained, the first target text and the second target text are set as the second candidate text.
  • the convergent node text of the keyword is set as the user's attention node text
  • the third candidate text outside the first candidate text is selected from the text updated in the preset text database within a preset time period, and the third candidate text is selected through the preset word segmentation tool Perform word segmentation processing on the title of each text in the third candidate text to obtain the title vector of each text in the third candidate text; calculate the distance between the title vector and the node vector of each convergent node text in the affair map
  • the second node distance; the second node distance selected from the third candidate text is less than the second preset distance, and the convergent node text corresponding to the less than the second preset distance is the first target text of the user-focused node text , Or select the second node distance from the third candidate
  • the present application provides another embodiment of a text recommendation method.
  • a text recommendation method according to the operation behavior, from the first candidate text and the second candidate
  • the steps of filtering out the selected text from the text and recommending the selected text to the target user include:
  • Step S41 Obtain the spread of each text in the first candidate text and the second candidate text, and obtain the correlation degree of each text with the target user, and obtain the target user's information according to the operation behavior. Preference
  • the calculation steps of the dissemination amount can be as follows: first delete the titles of the first candidate text and the second candidate text (Due to the text collection process, some punctuation marks may be changed from half-width to full-width. In addition, some media will also modify some punctuation marks from half-width to full-width or from full-width when forwarding text.
  • the step of obtaining the relevance of each text to the target user includes:
  • Step S41 Obtain the number of times the keyword appears in each text of the first candidate text, and set the number of times as the number of words;
  • Step S42 Obtain the position where the keyword appears in each text of the first candidate text, set the position as a word position, and obtain a preset position weight corresponding to the word position, where the word position The position weight is different.
  • the word position includes the first sentence position of the first paragraph of the text, the first sentence position of the end paragraph of the text, the position of the first sentence of the text, the position of the non-first sentence of the end of the text, the position of the first sentence of the non-first paragraph, and the position of the non-end The position of the first sentence of the paragraph;
  • Step S43 Obtain the ratio of the number of sentences between the first and last occurrences of the keyword in each text of the first candidate text to the total number of sentences in the full text, and set the ratio as the word span;
  • Step S44 Obtain the target text between the first and last occurrences of the keyword in each text of the first candidate text, and obtain the average number of sentences contained in each preset sentence in the target text.
  • the number of said keywords, the average number of said keywords contained in each preset sentence is set as the word density;
  • Step S45 obtaining the first degree of relevance of each text in the first candidate text according to the number of words, the preset position weight corresponding to the word position, the word span and the word density;
  • the calculation of the correlation between the first candidate text and the second candidate text is different, as shown in FIG. 13.
  • the first degree of relevance of the first candidate text is calculated as follows: the number of times, word positions, word spans, word density, etc. of the keywords appearing in each text of the first candidate text are obtained, according to the word times and word positions , Word span, word density to obtain the first degree of relevance of each text of the first candidate text, specifically, the number of words a: the total number of keywords appearing in the text body; word position b: the initial value of b The value is 0, if the keyword appears in the first sentence of the first paragraph or the first sentence of the last paragraph of the text body, add 2 to b; if the keyword appears in the first sentence or the last sentence of the text body, it will be b plus 1; if the keyword appears in the first sentence of the remaining paragraphs except the first and last paragraphs, add 0.5 to b; word span c: between the first and last occurrence of the keyword in the text The ratio of the number of sentences in the interval to the total number of sentences in the full text; word density d: intercept the text between the first and last occurrence
  • Step S46 Obtain the screening logic depth of each text in the second candidate text, and determine the second relevance of each text in the second candidate text according to the screening logic depth.
  • the step of obtaining the preference degree of the target user according to the operation behavior includes:
  • Step S47 Obtain the historical browsing text of the target user from the operation behavior, obtain the first document vector of each text in the historical browsing text, and obtain the first candidate text and the second candidate text The second document vector of each text in the middle;
  • Step S48 Obtain a first Pearson correlation coefficient between the second document vector and the first document vector, and acquire the preference degree of the target user according to the first Pearson correlation coefficient.
  • the historical browsing text of the target user text is obtained.
  • the historical browsing text may be the historical browsing text in the past month.
  • the first document vector of each text in the historical browsing text is obtained, and the The second document vector of each text in the first candidate text and the second candidate text is obtained, the first Pearson correlation coefficient between the second document vector and the first document vector is obtained, and the first Pearson correlation coefficient is obtained according to the first document vector.
  • the Pearson correlation coefficient obtains the preference degree of the target user, and obtains the user preference degree according to the historical browsing text and the first candidate text and the second candidate text.
  • Step S42 According to the amount of communication, the degree of relevance and the degree of preference, the selected text is selected from the first candidate text and the second candidate text, and the selected text is recommended to all The target user.
  • the amount of communication, the degree of relevance, and the degree of preference are integrated, the selected text is selected from the first candidate text and the second candidate text, and the selected text is recommended To the target user.
  • the information is obtained according to the operation behavior.
  • the preference degree of the target user according to the communication volume, the correlation degree and the preference degree, the selected text is selected from the first candidate text and the second candidate text, and the selected text is selected
  • the text is recommended to the target user.
  • three factors are considered to filter the selected text, so as to improve the accuracy of recommendation.
  • this application provides another embodiment of a text recommendation method.
  • the step of obtaining the first document vector of each text in the historical browsing text includes:
  • Step B1 Acquire a first probability matrix in which each text in the historically browsed text is classified under a first preset category according to a preset clustering algorithm
  • Step B2 Obtain the word segmentation words of each text in the historical browsing text according to a preset word segmentation algorithm, and obtain a second probability matrix in which the word segmentation words are classified into a second preset category;
  • Step B3 Obtain each optimized word vector corresponding to each text in the historically browsed text according to the first probability matrix and the second probability matrix;
  • Step B4 Obtain the first document vector of each text in the historically browsed text according to the optimized word vector.
  • the first probability matrix of each text in the historical browsing text is divided into the first preset category (including the number of text subcategories).
  • LDA Topic Dirichlet Allocation
  • Implied Dirichlet distribution for unsupervised clustering of historically browsed texts (the number of clusters can be set to 200) to obtain the first probability matrix p for each text to be classified into the first preset category
  • Obtain each optimized word vector W corresponding to each text in the historical browsing text according to the first probability matrix and the second probability matrix, W 0.6p+0.4q, and obtain the history according to the optimized word vector Browse the first document vector of each text in the text.
  • the optimized word vectors of all word segments are added to obtain the document vector of the corresponding news.
  • the first Pearson correlation coefficient between the second document vector and the first document vector is acquired, and the preference of the target user is acquired according to the first Pearson correlation coefficient, firstly collecting from a preset text database Retrieve the text collection (N 1 , N2,...Nk) that the current target user has clicked and browsed in the history, a total of k texts. Then, the first candidate text, the second candidate text, and the retrieved k texts such as news are respectively subjected to preset stuttering word segmentation processing, and the word vectors of all the word segments are added to obtain the document vector V of the corresponding text.
  • V 1 ,V2,...Vk Use (V 1 ,V2,...Vk) to represent the document vector of each text in the historically browsed text collection (N 1 ,N2,...Nk), and use Vv to represent the current first candidate text
  • ⁇ (a,b) to represent the first Pearson correlation coefficient between the two document vectors (a,b)
  • the user preference of the second candidate text can be expressed as
  • Vv in the formula represents the document vector of each text in the first candidate text and the second candidate text
  • V j represents the document vector of each text in the historically browsed text collection.
  • the step of obtaining the first Pearson correlation coefficient between the second document vector and the first document vector, and obtaining the preference degree of the target user according to the first Pearson correlation coefficient includes:
  • Step C1 obtaining the historical browsing time from the time when each text in the historical browsing text is clicked to browse to the current moment;
  • Step C2 Obtain the first Pearson correlation coefficient between the second document vector and the first document vector, and perform interest weight reduction processing on the first Pearson correlation coefficient according to the historical browsing time to obtain the first Pearson correlation coefficient.
  • Step C3 Obtain the preference degree of the target user according to the second Pearson correlation coefficient.
  • the text that a certain operator is most concerned about 1 week ago is the company's new product launch
  • the current text that is most concerned about is the public's evaluation of the company's new products. Therefore, it is necessary to perform time interest reduction processing on the text that the user has clicked and browsed in the history. First, obtain the historical browsing time from the time each text in the historical browsing text is clicked to the current moment.
  • use t k indicates that the text N k was clicked to browse before t k days, that is, the historical browsing time is t k , the first Pearson correlation coefficient between the second document vector and the first document vector is obtained, and according to the Historical browsing time performs interest reduction processing on the first Pearson correlation coefficient to obtain a second Pearson correlation coefficient, and obtains the preference of the target user according to the second Pearson correlation coefficient, then the first candidate text and The user preference of the second candidate text is finally expressed as
  • this application provides another embodiment of a text recommendation method.
  • the degree of relevance and the degree of preference from the The step of screening the selected text from the first candidate text and the second candidate text, and recommending the selected text to the target user includes:
  • Step D1 Calculate the value score of each of the first candidate text and the second candidate text according to the communication volume, the first degree of relevance, the second degree of relevance, and the preference degree;
  • step D2 a preset number of texts are selected as selected texts in order from high to low according to the value score, and the selected texts are recommended to the target user.
  • the first degree of relevance, the second degree of relevance, and the user preference degree selected text is selected from the first candidate text and the second candidate text, and the selected text is selected from among the first candidate text and the second candidate text.
  • the selected text is pushed to the target user.
  • a preset number of texts are selected as selected texts in sequence at the lowest level, and the selected texts are recommended to the target user. For example, the 10 news with the highest scores are selected as the target content and pushed to the target user, which can be pushed once a day.
  • the value score by calculating each of the first candidate text and the second candidate text according to the amount of communication, the first degree of relevance, the second degree of relevance, and the preference degree The value score; according to the value score, a preset number of texts are selected in order from high to low as the selected text, and the selected text is recommended to the target user. In this embodiment, accurate text recommendation is performed based on the value score.
  • FIG. 3 is a schematic diagram of the device structure of the hardware operating environment involved in the solution of the embodiment of the present application.
  • the text recommendation device in the embodiment of the present application may be a PC, or a terminal device such as a smart phone, a tablet computer, and a portable computer.
  • the text recommendation device may include a processor 1001, such as a CPU, a memory 1005, and a communication bus 1002.
  • the communication bus 1002 is used to implement connection and communication between the processor 1001 and the memory 1005.
  • the memory 1005 may be a high-speed RAM memory, or a non-volatile memory (non-volatile memory), such as a magnetic disk memory.
  • the memory 1005 may also be a storage device independent of the aforementioned processor 1001.
  • the text recommendation device may further include a target user interface, a network interface, a camera, an RF (Radio Frequency) circuit, a sensor, an audio circuit, a WiFi module, and so on.
  • the target user interface may include a display screen (Display) and an input unit such as a keyboard (Keyboard), and the optional target user interface may also include a standard wired interface and a wireless interface.
  • the network interface can optionally include a standard wired interface and a wireless interface (such as a WI-FI interface).
  • the structure of the text recommendation device shown in FIG. 3 does not constitute a limitation on the text recommendation device, and may include more or less components than those shown in the figure, or a combination of certain components, or different components Layout.
  • the memory 1005 as a computer storage medium may include an operating system, a network communication module, and a text recommendation program.
  • the operating system is a program that manages and controls the hardware and software resources of the text recommendation device, and supports the operation of the text recommendation program and other software and/or programs.
  • the network communication module is used to implement communication between various components in the memory 1005 and communication with other hardware and software in the text recommendation device.
  • the processor 1001 is configured to execute the text recommendation program stored in the memory 1005 to implement the steps of any one of the text recommendation methods described above.
  • an embodiment of the present application also proposes a text recommendation device, and the text recommendation device includes:
  • the monitoring module is used to monitor the operation behavior of the target user, and determine the keywords associated with the target user according to the operation behavior;
  • a retrieval module configured to retrieve more than one updated text containing at least one of the keywords from a preset text database set as the first candidate text;
  • the selection module is configured to retrieve a preset affair atlas, and select from the preset text database set according to the preset affair atlas with a total relevance to the first candidate text that is not less than a preset relevance threshold Update the text, as the second candidate text, the affair map includes the association relationship between the text and the text, and each association relationship has its corresponding degree of association;
  • the screening module is configured to screen out the selected text from the first candidate text and the second candidate text according to the operation behavior, and recommend the selected text to the target user.
  • the text recommendation device further includes:
  • the collection module is configured to collect the text to be processed from the preset text database collection every preset time period;
  • the preprocessing module is used to perform html tag filtering, symbol filtering, and sentence processing on the to-be-processed text through a preset regular expression to obtain a preprocessed text composed of a sentence list;
  • the generating module is used to generate the preset affair atlas according to the preprocessed text.
  • the generating module includes:
  • the recognition unit is configured to recognize multiple preset text association relationships for each clause in the clause list to obtain the node text to be processed, wherein the preset text association relationships include, but are not limited to, Shun Cheng, Causality, conditions, and parallel relationships;
  • the first obtaining unit is configured to perform word segmentation processing on the node text to be processed by a preset word segmentation tool, and obtain the word vector of each word segmentation, and obtain the node vector of each node text to be processed based on the word vector of each word segmentation;
  • the first calculation unit is configured to calculate the first node distance between each to-be-processed node text and other to-be-processed node texts according to the node vector of each to-be-processed node text;
  • the grafting processing unit is configured to perform iterative grafting processing on two to-be-processed node texts whose node distance is less than a first preset distance, until each of the to-be-processed node texts is in a convergent state where the node text relationship edge no longer changes, wherein , Setting each of the node texts to be processed in a convergent state as the convergent node text;
  • the generating unit is configured to generate the preset affair graph based on the node text relationship edge between the convergent node text and the convergent node text.
  • the selection module includes:
  • the retrieval unit is configured to retrieve a preset affair map, and determine whether there is a convergent node text in the corresponding clause that contains the keyword in the affair map;
  • the first setting unit is configured to, if it exists, set the convergent node text containing the keyword in the corresponding clause as the user-focused node text, and update it from the preset text database within a preset time period Select a third candidate text other than the first candidate text from the text, and perform word segmentation processing on the title of each text in the third candidate text by a preset word segmentation tool to obtain a title vector of each text in the third candidate text;
  • a second calculation unit configured to calculate the second node distance between the title vector and the node vector of each convergent node text in the affair map
  • the selecting unit is configured to select, from the third candidate text, a second node distance less than a second preset distance, and the convergent node text corresponding to the second preset distance is the first target text of the node text that the user pays attention to; Or select the second node distance from the third candidate text to be less than the second preset distance, and the preset filtering logic depth range of the convergent node text corresponding to the second preset distance is less than the second node text of the user concerned.
  • Target text wherein the logical depth of the screening is determined according to the degree of relevance of the respective association relationships in the affair map;
  • the second setting unit is configured to set the first target text and the second target text as the second candidate text.
  • the screening module includes:
  • the second acquiring unit is configured to acquire the spread of each text in the first candidate text and the second candidate text, and acquire the relevance of each text to the target user, and acquire the information according to the operation behavior. State the preference of the target user;
  • a recommendation unit configured to filter out selected text from the first candidate text and the second candidate text according to the amount of communication, the degree of relevance, and the degree of preference, and recommend the selected text To the target user.
  • the second acquiring unit includes:
  • the first obtaining subunit is configured to obtain the number of times the keyword appears in each text of the first candidate text, and set the number of times as the number of words;
  • the second obtaining subunit is used to obtain the position where the keyword appears in each text of the first candidate text, set the position as a word position, and obtain a preset position weight corresponding to the word position , Where the word position is different, the position weight is different, the word position includes the first sentence position of the first paragraph of the text, the first sentence position of the last paragraph of the text, the position of the first sentence of the text, the position of the last sentence of the text, the position of the non-first paragraph Sentence position and the position of the first sentence of the non-final paragraph
  • the third acquisition subunit is used to acquire the ratio of the number of sentences between the first and last occurrences of the keyword in each text of the first candidate text to the total number of sentences in the full text, and compare all The ratio value is set as the word span;
  • the fourth obtaining subunit is used to obtain the target text between the first and last occurrences of the keyword in each text of the first candidate text, and obtain the average preset value in the target text.
  • the number of sentences contains the number of the keywords, and the average number of the keywords contained in each preset number of sentences is set as the word density;
  • the fifth obtaining subunit is configured to obtain the first degree of relevance of each text in the first candidate text according to the number of words, the preset position weight corresponding to the word position, the word span and the word density ;
  • the sixth obtaining subunit is configured to obtain the screening logic depth of each text in the second candidate text, and determine the second relevance of each text in the second candidate text according to the screening logic depth.
  • the second acquiring unit includes:
  • the seventh obtaining subunit is configured to obtain the historical browsing text of the target user from the operation behavior, obtain the first document vector of each text in the historical browsing text, and obtain the first candidate text and all the texts. State the second document vector of each text in the second candidate text;
  • the eighth obtaining subunit is configured to obtain the first Pearson correlation coefficient between the second document vector and the first document vector, and obtain the preference degree of the target user according to the first Pearson correlation coefficient.
  • the seventh acquiring subunit is used to implement:
  • the eighth acquiring subunit is used to implement:
  • the screening module includes:
  • the third calculation unit is configured to calculate each of the first candidate text and the second candidate text according to the amount of communication, the first degree of relevance, the second degree of relevance, and the preference degree Value score
  • the screening unit is configured to select a preset number of texts as selected texts in order from high to low according to the value score, and recommend the selected texts to the target user.
  • the specific implementation of the text recommendation device is basically the same as the above embodiments of the text recommendation method, and will not be repeated here.
  • an embodiment of the present application also proposes a text recommendation device.
  • the device includes a memory 109, a processor 110, and a text recommendation program stored on the memory 109 and running on the processor 110, and the text recommendation program is executed by the processor 110.
  • this application also provides a computer medium that stores one or more programs, and the one or more programs may also be executed by one or more processors to implement the above text recommendation method Steps of each embodiment.
  • the technical solution of this application essentially or the part that contributes to the existing technology can be embodied in the form of a software product, and the computer software product is stored in a storage medium (such as ROM/RAM, magnetic disk, The optical disc) includes several instructions to make a terminal (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) execute the methods described in the various embodiments of the present application.
  • a terminal which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.

Landscapes

  • Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Animal Behavior & Ethology (AREA)
  • Computational Linguistics (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

L'invention concerne un procédé, un appareil et un dispositif de recommandation de texte, ainsi qu'un support, se rapportant au domaine technique de la technologie financière. Ledit procédé comprend les étapes suivantes : surveillance d'un comportement opérationnel d'un utilisateur cible, de façon à déterminer des mots-clés associés à l'utilisateur cible ; recherche dans un ensemble de bases de données de texte prédéfini d'un ou de plusieurs textes de mise à jour contenant au moins l'un des mots-clés en tant que premier texte candidat ; invocation d'un graphe logique d'événement prédéfini et, selon le graphe, sélection, à partir de l'ensemble de bases de données de texte prédéfini, d'un texte de mise à jour dont le degré d'association totale avec le premier texte candidat n'est pas inférieur à un seuil d'association prédéfini, et adoption de celui-ci en tant que deuxième texte candidat, le graphe logique d'événement contenant une relation d'association entre les textes et chaque relation d'association ayant un degré d'association lui correspondant ; et en fonction du comportement opérationnel, filtrage d'un texte sélectionné parmi le premier texte candidat et le deuxième texte candidat, et recommandation de celui-ci à l'utilisateur cible.
PCT/CN2020/129115 2019-11-22 2020-11-16 Procédé, appareil et dispositif de recommandation de texte, et support WO2021098648A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201911179808.XA CN110888990B (zh) 2019-11-22 2019-11-22 文本推荐方法、装置、设备及介质
CN201911179808.X 2019-11-22

Publications (1)

Publication Number Publication Date
WO2021098648A1 true WO2021098648A1 (fr) 2021-05-27

Family

ID=69748961

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/129115 WO2021098648A1 (fr) 2019-11-22 2020-11-16 Procédé, appareil et dispositif de recommandation de texte, et support

Country Status (2)

Country Link
CN (1) CN110888990B (fr)
WO (1) WO2021098648A1 (fr)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114020936A (zh) * 2022-01-06 2022-02-08 北京融信数联科技有限公司 多模态事理图谱的构建方法、系统和可读存储介质
CN114817678A (zh) * 2022-01-27 2022-07-29 武汉理工大学 一种面向特定领域的自动文本采集方法
CN118194864A (zh) * 2024-05-17 2024-06-14 上海通创信息技术股份有限公司 一种基于语音分析的潜在用户挖掘方法及系统

Families Citing this family (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110888990B (zh) * 2019-11-22 2024-04-12 深圳前海微众银行股份有限公司 文本推荐方法、装置、设备及介质
CN111428092B (zh) * 2020-03-20 2023-05-02 北京中亦安图科技股份有限公司 基于图模型的银行精准营销方法
CN111400456B (zh) * 2020-03-20 2023-09-26 北京百度网讯科技有限公司 资讯推荐方法及装置
CN112000795A (zh) * 2020-08-04 2020-11-27 中国建设银行股份有限公司 公文推荐方法和装置
CN112561581A (zh) * 2020-12-14 2021-03-26 珠海格力电器股份有限公司 一种推荐方法、装置、电子设备及存储介质
CN112836061A (zh) * 2021-01-12 2021-05-25 平安科技（深圳）有限公司 智能推荐的方法、装置以及计算机设备
CN112749344B (zh) * 2021-02-04 2023-08-01 北京百度网讯科技有限公司 信息推荐方法、装置、电子设备、存储介质及程序产品
CN113505587B (zh) * 2021-06-23 2024-04-09 科大讯飞华南人工智能研究院（广州）有限公司 实体抽取方法及相关装置、设备和存储介质
US11977841B2 (en) 2021-12-22 2024-05-07 Bank Of America Corporation Classification of documents
CN114625747B (zh) * 2022-05-13 2022-08-12 杭银消费金融股份有限公司 基于信息安全的风控更新方法及系统

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150310073A1 (en) * 2014-04-29 2015-10-29 Microsoft Corporation Finding patterns in a knowledge base to compose table answers
CN109033132A (zh) * 2018-06-05 2018-12-18 中证征信（深圳）有限公司 利用知识图谱计算文本和主体相关度的方法以及装置
CN109165350A (zh) * 2018-08-23 2019-01-08 成都品果科技有限公司 一种基于深度知识感知的信息推荐方法和系统
CN110888990A (zh) * 2019-11-22 2020-03-17 深圳前海微众银行股份有限公司 文本推荐方法、装置、设备及介质

Family Cites Families (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104239298B (zh) * 2013-06-06 2018-10-30 腾讯科技（深圳）有限公司 文本信息推荐方法、服务器、浏览器及系统
CN105335519B (zh) * 2015-11-18 2021-08-17 百度在线网络技术（北京）有限公司 模型生成方法及装置、推荐方法及装置
US20170139899A1 (en) * 2015-11-18 2017-05-18 Le Holdings (Beijing) Co., Ltd. Keyword extraction method and electronic device
CN108733694B (zh) * 2017-04-18 2022-02-25 北京国双科技有限公司 检索推荐方法和装置
CN107944911B (zh) * 2017-11-18 2021-12-03 电子科技大学 一种基于文本分析的推荐系统的推荐方法
CN108153901B (zh) * 2018-01-16 2022-04-19 北京百度网讯科技有限公司 基于知识图谱的信息推送方法和装置
CN109408826A (zh) * 2018-11-07 2019-03-01 北京锐安科技有限公司 一种文本信息提取方法、装置、服务器及存储介质
CN109597878B (zh) * 2018-11-13 2020-06-05 北京合享智慧科技有限公司 一种确定文本相似度的方法及相关装置
CN110413875B (zh) * 2019-06-26 2024-06-07 腾讯科技（深圳）有限公司 一种文本信息推送的方法以及相关装置

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150310073A1 (en) * 2014-04-29 2015-10-29 Microsoft Corporation Finding patterns in a knowledge base to compose table answers
CN109033132A (zh) * 2018-06-05 2018-12-18 中证征信（深圳）有限公司 利用知识图谱计算文本和主体相关度的方法以及装置
CN109165350A (zh) * 2018-08-23 2019-01-08 成都品果科技有限公司 一种基于深度知识感知的信息推荐方法和系统
CN110888990A (zh) * 2019-11-22 2020-03-17 深圳前海微众银行股份有限公司 文本推荐方法、装置、设备及介质

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114020936A (zh) * 2022-01-06 2022-02-08 北京融信数联科技有限公司 多模态事理图谱的构建方法、系统和可读存储介质
CN114817678A (zh) * 2022-01-27 2022-07-29 武汉理工大学 一种面向特定领域的自动文本采集方法
CN118194864A (zh) * 2024-05-17 2024-06-14 上海通创信息技术股份有限公司 一种基于语音分析的潜在用户挖掘方法及系统

Also Published As

Publication number Publication date
CN110888990A (zh) 2020-03-17
CN110888990B (zh) 2024-04-12

Similar Documents

Publication Publication Date Title
WO2021098648A1 (fr) Procédé, appareil et dispositif de recommandation de texte, et support
US11580104B2 (en) Method, apparatus, device, and storage medium for intention recommendation
US11663254B2 (en) System and engine for seeded clustering of news events
US10217058B2 (en) Predicting interesting things and concepts in content
US7912816B2 (en) Adaptive archive data management
WO2018072071A1 (fr) Système et procédé de construction de carte de connaissances
WO2017024884A1 (fr) Procédé et dispositif d'identification d'intention de recherche
US20130060769A1 (en) System and method for identifying social media interactions
WO2018040068A1 (fr) Système et procédé d'analyse sémantique sur la base d'un graphique de connaissances
US10839013B1 (en) Generating a graphical representation of relationships among a set of articles and information associated with the set of articles
US10002187B2 (en) Method and system for performing topic creation for social data
CN110232126B (zh) 热点挖掘方法及服务器和计算机可读存储介质
US20140201203A1 (en) System, method and device for providing an automated electronic researcher
CN113297457B (zh) 一种高精准性的信息资源智能推送系统及推送方法
US20210256221A1 (en) System and method for automatic summarization of content with event based analysis
CN112825089A (zh) 文章推荐方法、装置、设备及存储介质
US8799314B2 (en) System and method for managing information map
WO2015084757A1 (fr) Systèmes et procédés de traitement de données stockées dans une base de données
KR20190109628A (ko) 개인화된 기사 컨텐츠 제공 방법 및 장치
US9323721B1 (en) Quotation identification
AU2018273369A1 (en) Automated classification of network-accessible content
US20230090601A1 (en) System and method for polarity analysis
KR20130082879A (ko) 소셜네트워크서비스 게시글의 메타정보에 기반한 음악매칭을 통한 통합적 음악 서비스 제공 방법
CN108520012A (zh) 基于机器学习的移动互联网用户评论挖掘方法
US8195458B2 (en) Open class noun classification

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20890760

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20890760

Country of ref document: EP

Kind code of ref document: A1