CN111914079A - Topic recommendation method and system based on user tags - Google Patents
Topic recommendation method and system based on user tags Download PDFInfo
- Publication number
- CN111914079A CN111914079A CN202010787292.3A CN202010787292A CN111914079A CN 111914079 A CN111914079 A CN 111914079A CN 202010787292 A CN202010787292 A CN 202010787292A CN 111914079 A CN111914079 A CN 111914079A
- Authority
- CN
- China
- Prior art keywords
- user
- topic
- topics
- keyword
- target
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
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
- G06F16/353—Clustering; Classification into predefined classes
-
- 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 application discloses a topic recommendation method and system based on user tags; the method comprises the following steps: selecting topic keywords from a keyword tag library in the cloud; different keywords and frequency numbers are stored in the keyword tag library in a classified manner; inputting the topic keywords into the trained texts in different modes to generate a neural network model, and obtaining the topics in different modes of the topic keywords; and acquiring the reading label of the user, searching the topic matched with the reading label of the user according to the reading label of the user, and recommending the topic to the user. By the method and the device, accurate pushing can be achieved according to the user tags, topics which are interesting to the user are recommended, and the possibility that the user participates in topic discussion is higher.
Description
Technical Field
The application relates to the field of data processing, in particular to a topic recommendation method and system based on a user tag.
Background
With the increasing popularization of the internet, the internet forum and the network community in the professional field become important channels for people to acquire and exchange professional information. A large number of netizens publish own opinion insights, release distribution industry trends, research and develop new progresses or breakthroughs, hotspot science and technology news and the like in a network community, and in order to improve the liveness of the network community, some hotspot topics are often found out from various opinions, the hotspot news and the like so as to be discussed and communicated by users. If the topic provided for the user is good and interesting, the enthusiasm of the user for participating in the topic discussion can be greatly improved, and the activity of the network community is improved.
The existing topic pushing mode is generally to directly display a hot topic of a network community on a home page or push the hot topic to a user, and the hot topic is not necessarily interesting for each user. Therefore, the pushing mode has certain limitation, the pushing effect is limited, the attached topics cannot be pushed according to the individuation of the user, and the topic participation degree and the user experience are improved.
Disclosure of Invention
In order to solve the technical problems, the application provides a topic recommendation method and system based on a user tag, and specifically, the technical scheme of the application is as follows:
on one hand, the application discloses a topic recommendation method based on a user tag, which comprises the following steps: selecting topic keywords from a keyword tag library in the cloud; different keywords and frequency numbers are stored in the keyword tag library in a classified manner; inputting the topic keywords into trained texts in different modes to generate a neural network model, and obtaining topics in different modes of the topic keywords; the method comprises the steps of obtaining a reading label of a user, searching a topic matched with the reading label of the user according to the reading label of the user, and recommending the topic to the user.
Preferably, the topic recommendation method based on the user tag further includes: acquiring historical data information of searching and reading of the user; extracting keywords from the historical data information, and counting the frequency of each keyword; acquiring the text mode preference of the user according to the historical data information; and determining the reading label of the user according to the counted keywords, the counted frequency and the preference of the text mode.
Preferably, the user's reading tag includes: reading keywords and frequency thereof, and text mode preference; the searching for the topic matched with the reading label of the user according to the reading label of the user and recommending the topic to the user specifically comprises: acquiring a keyword with the highest frequency number in the reading label of the user as a target keyword; searching topics matched with the target keywords; when the topics matched with the target keywords are found, selecting the topics matched with the text mode preference in the reading label of the user as target topics from the found topics; recommending the target topic to the user.
Preferably, the user tag-based topic recommendation method further includes: training texts in different modes to generate a neural network model; the method specifically comprises the following steps: crawling hot topics with reading quantity exceeding a preset threshold from the internet; classifying the collected hot topics according to a text mode; extracting keywords in the hot topics, and labeling to obtain training samples; inputting the training samples into different neural network models according to the types of the text modes for iterative training, and obtaining text generation neural network models of different modes.
Preferably, the user tag-based topic recommendation method further includes: when the topic matched with the reading label of the user is not found, determining a target keyword and a text mode according to the reading label of the user; selecting a text generation neural network model of a corresponding mode as a target text generation neural network model according to the determined text mode; inputting the determined target keywords into the target text to generate a neural network model, and generating a target topic in a corresponding mode; recommending the target topic to the user.
Preferably, the user tag-based topic recommendation method further includes: when the topic matched with the reading label of the user is not found, sequencing all topics in the website according to the reading amount or participation amount to obtain a plurality of hot topics; recommending the number of hot topics to the user.
Preferably, the keyword tag library further stores a keyword extension map; after the keyword with the highest frequency number in the reading label of the user is obtained as the target keyword, the method comprises the following steps: searching for an extension keyword of the target keyword according to the keyword extension map; searching topics matched with the target keywords or the extension keywords; when topics matched with the target keywords or the extension keywords are found, selecting topics matched with the text mode preference in the reading label of the user from the found topics as target topics; recommending the target topic to the user.
Preferably, after obtaining the topics of the different patterns of the topic keywords, the method further comprises: searching topics relevant to the keywords from published topics as alternative topics; and screening out topics with similarity reaching a preset threshold value with the alternative topics from the topics acquired through the neural network model, and deleting the topics.
On the other hand, the application also discloses a topic recommendation system based on the user tags, which comprises: the keyword tag library is used for storing different keywords and frequency numbers in a classified manner; the keyword tag library is established in the cloud; the keyword selection module is used for selecting topic keywords from a keyword tag library in the cloud; the topic generation module is used for inputting the topic keywords into trained texts in different modes to generate a neural network model so as to obtain topics in different modes of the topic keywords; the tag acquisition module is used for acquiring a reading tag of a user; and the topic recommendation module is used for searching topics matched with the reading labels of the users according to the reading labels of the users and recommending the topics to the users.
Preferably, the tag obtaining module includes: the historical data acquisition submodule is used for acquiring searching and reading historical data information of the user; the extraction and statistics submodule is used for extracting keywords from the historical data information and counting the frequency of each keyword; the data analysis submodule is used for acquiring the text mode preference of the user according to the historical data information; and determining the reading label of the user according to the counted keywords, the frequency and the preference of the text mode.
Preferably, the user's reading tag includes: reading keywords and frequency thereof, and text mode preference; the topic recommendation module specifically comprises: the target acquisition submodule is used for acquiring a keyword with the highest frequency number in the reading label of the user as a target keyword; the matching search sub-module is used for searching topics matched with the target keywords; the topic selection submodule is used for selecting topics matched with the text mode preference in the reading label of the user as target topics from the searched topics when the topics matched with the target keywords are searched; and the information recommendation submodule is used for recommending the target topic to the user.
Preferably, the user tag-based topic recommendation system further comprises: the model training module is used for training texts in different modes to generate a neural network model; the model training module specifically comprises: the information crawling sub-module is used for crawling hot topics with reading quantity exceeding a preset threshold value from the internet; the information processing submodule is used for classifying the collected hot topics according to the text mode; extracting keywords in the hot topics for labeling to obtain training samples; and the iterative training submodule is used for inputting the training samples into different neural network models according to the types of the text modes for iterative training to obtain the text generation neural network models of different modes.
Preferably, the topic recommendation module further comprises: a model selection submodule, wherein: the target obtaining sub-module is further used for determining a target keyword and a text mode according to the reading label of the user when the matching searching sub-module does not find the topic matched with the target keyword; the model selection submodule is used for selecting a text generation neural network model of a corresponding mode as a target text generation neural network model according to the determined text mode; inputting the determined target keywords into the target text through the topic generation module to generate a neural network model, and generating a target topic in a corresponding mode; the information recommending submodule is also used for recommending the target topic to the user.
Preferably, the user tag-based topic recommendation system further includes: the topic sequencing module is used for sequencing all topics in the website according to the reading amount or participation amount when the topics matched with the reading tags of the users are not found, and obtaining a plurality of hot topics; the topic recommending module is further used for recommending the hot topics to the user.
Preferably, the keyword tag library further stores a keyword extension map; the topic recommendation module further comprises: an association search submodule; wherein: the correlation searching sub-module is used for searching an extension keyword of the target keyword according to the keyword extension map; the matching search sub-module is also used for searching topics matched with the target keywords or the extension keywords; the topic selection sub-module is further configured to select, from the searched topics, a topic that matches the text mode preference in the reading tag of the user as a target topic when the topics that match the target keyword or the extension keyword are found.
The application at least comprises the following technical effects:
(1) according to the method and the device, topic keywords are selected from the keyword tag library, then the neural network models are generated by using texts in different modes to generate topics in different types, then matching is carried out according to reading tags of users, and topics matched with the user tags (such as topics in pen-and-ink modes which are possibly interesting to the users) are selected to be recommended, so that the possibility that the users participate in discussion of the topics is greatly improved, and the activity of a network community is further improved.
(2) When the topic recommendation is carried out, besides the topic contents of keywords which are possibly interested by the user, the preference of the user to the text mode is further considered, and the found topic which is related to the interested keywords and is in the favorite text mode is used as the target topic, so that not only the content of the target keyword is interested by the user, but also the favorite stroke method of the user is obtained, the favor of the user is obtained, the user is easy to touch, and the user is more willing to participate in the discussion of the topic.
(3) According to the method and the device, not only can the target speech be generated or selected according to the selected target keyword, but also the extension keyword of the target keyword can be obtained according to the keyword extension map stored in the keyword tag library, and then the corresponding topic related to the extension keyword is obtained through a text generation neural network model according to the extension keyword. Since the extension keyword is an extension of and related to the target keyword, if the user is interested in the topic of the target keyword, the user is also likely to be interested in the extension topic of the target keyword. Therefore, when the topics are recommended to the user, besides the topics of the direct target keywords, the topics of the extension keywords related to the target keywords can be recommended, so that the user can participate in a plurality of different interesting topics, and the topic recommendation range is widened on the premise of ensuring the recommendation accuracy.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without inventive exercise.
FIG. 1 is a flowchart of an embodiment of a user tag-based topic recommendation method according to the present application;
FIG. 2 is a flowchart illustrating another embodiment of the topic recommendation method based on user tags according to the present application;
FIG. 3 is a flowchart illustrating another embodiment of the topic recommendation method based on user tags according to the present application;
FIG. 4 is a block diagram of an embodiment of the user tag based topic recommendation system of the present application;
FIG. 5 is a block diagram of an embodiment of the user tag based topic recommendation system of the present application;
fig. 6 is a block diagram illustrating an embodiment of the user tag-based topic recommendation system according to the present application.
Detailed Description
In the following description, for purposes of explanation and not limitation, specific details are set forth, such as particular system structures, techniques, etc. in order to provide a thorough understanding of the embodiments of the present application. However, it will be apparent to one skilled in the art that the present application may be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present application with unnecessary detail.
It will be understood that the terms "comprises" and/or "comprising," when used in this specification and the appended claims, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.
For the sake of simplicity, the drawings only schematically show the parts relevant to the present application, and they do not represent the actual structure of the product. In addition, in order to make the drawings concise and understandable, components having the same structure or function in some of the drawings are only schematically depicted, or only one of them is labeled. In this document, "one" means not only "only one" but also a case of "more than one".
It should be further understood that the term "and/or" as used in this specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.
In addition, in the description of the present application, the terms "first", "second", and the like are used only for distinguishing the description, and are not intended to indicate or imply relative importance.
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following description will be made with reference to the accompanying drawings. It is obvious that the drawings in the following description are only some examples of the present application, and that for a person skilled in the art, other drawings and other embodiments can be obtained from these drawings without inventive effort.
The application discloses a topic recommendation method based on a user tag, wherein the first embodiment is shown in fig. 1 and comprises the following steps:
s101, selecting topic keywords from a keyword tag library in a cloud; different keywords and frequency numbers are stored in the keyword tag library in a classified manner;
specifically, the keyword tag library is arranged at the cloud end, so that a data multiple real-time storage technology can be provided, and data loss is avoided. The keyword tag library stores different keywords and the frequency numbers of the keywords according to categories. The frequency is the number of statistics for the keyword. Generally, every time an article, news or post and other data information are newly added in a network community, the data information is processed, and keywords of the data information are extracted; if the keyword exists in the keyword tag library in the cloud, directly updating the frequency number (adding one) of the keyword in the keyword tag library; if the keyword is not stored in the keyword tag library, the category of the keyword is determined, then the keyword is added under the corresponding category in the keyword tag library, and the frequency count of the keyword is recorded (the addition is 1).
Selecting topic keywords from a keyword tag library in the cloud; specifically, the topic keywords may be selected according to the frequency of each keyword stored therein, and a keyword with a high frequency indicates that the attention degree of the keyword in the network community is relatively high, so that the topic about the keyword is most likely to be a hot topic. Certainly, the topic keywords may be selected according to the frequency of the keywords, or the topic keywords may be selected according to the frequency increase rate, for example, some keywords, although the frequency is not very high, the frequency increase rate of the keywords in the last week is relatively fast, and then the related information of the keywords is likely to be a recent hotspot, so that the keywords whose frequency increase rate reaches a preset increase point in a preset time period may also be selected as the topic keywords; or selecting a plurality of keywords with frequency growth rate ranking at the top as topic keywords in a preset time period. Of course, the two manners may be combined, the embodiment does not limit the selection strategy of the topic keywords, and the corresponding selection strategy may be set according to the actual requirement to select the topic keywords that may become the hot spots from the numerous keywords in the keyword tag library. Of course, the topic keyword is not limited to only one keyword, and may include one or more keywords as a topic keyword together.
S102, inputting the topic keywords into trained texts in different modes to generate a neural network model, and obtaining topics in different modes of the topic keywords;
specifically, in the embodiment, different patterns of texts are trained to generate the neural network model, where the patterns refer to a style pattern or a writing method, and the same keyword and the same thing have different effects obtained by using different writing methods. For example, some people like a meta art, an emotional pattern; some people like a simple mode of short, concise and one-needle blood infusion; people like the glancing-pencil mode of witty humor; someone likes factual statements, a traditional news model. Specifically, different mode classes may be defined as desired. Inputting the same key words, the different patterns of text generation neural networks generate different patterns (different writing methods or different tones) of topics.
S103, acquiring a reading label of the user;
specifically, according to the historical reading data information of the user, the reading label of the user can be determined, and the user can know which kind of pen is preferred and which kind of data information is preferred.
S104, according to the reading label of the user, searching the topic matched with the reading label of the user, and recommending the topic to the user.
Specifically, after the reading label of the user is determined, the reading label can be searched and matched from various topics according to the reading label, and finally, the matched topic is recommended to the user to guide the user to participate in topic discussion. Matching topics, namely, matching is only carried out from the topics generated by the text generation neural network models; another implementation manner is that, in addition to the topics generated by the neural network model generated by each text according to the selected topic keywords in the embodiment, the topics may also be historical topics published before. The former has small matching workload; the latter match is more likely to succeed. Of course, matching may be performed from the generated topics first, and if the generated topics do not match, matching may be performed from the historical topics.
In the embodiment, topic keywords are selected from the keyword tag library, then the neural network models are generated by using texts in different modes to generate topics in different types, then matching is performed according to the reading tags of the users, and topics in pen modes which are possibly interesting to the users and are favored are selected to be recommended, so that the possibility that the users participate in the topic discussion is greatly improved, and the activity of the network community is further improved.
In the above embodiment, the user tag may be obtained as follows; specifically, in step S103 in the foregoing embodiment, the acquiring the reading tag of the user specifically includes:
s103-1, acquiring historical data information of searching and reading of the user;
s103-2, extracting keywords from the historical data information, and counting the frequency of each keyword;
s103-3, acquiring the preference of the text mode of the user according to the historical data information;
s103-4, determining the reading label of the user according to the counted keywords, the frequency and the preference of the text mode.
Specifically, by obtaining data information of historical search, browsing and reading of the user, keywords and frequency (corresponding to attention) of the user's attention can be roughly counted, for example, through data statistics, the user a pays attention to a keyword "XYZ" (medical expert name) in the medical expert type for 55 times; that is, the user a has searched or browsed up to 55 or up to 55 times the data information related to the expert "XYZ".
And according to the historical data information, the text mode preference of the user is acquired, the text type or the text mode of the data information browsed and read by the user needs to be analyzed, and the user likes the information of which text mode, for example, some like to see short and bold extremely-simplified text, and some like to have fine and smooth strokes, so that resonant emotion mode text is easily caused.
And finally, according to the counted keywords, the frequency of the keywords and the preference of the text mode, the reading label of the user can be determined. Specifically, the reading label determined in the scheme is not the general preference of the conventional determined user, the existing label for the user only provides a general preference range, for example, the label of a certain user B is a famous person, a diet health-preserving person and a cardiovascular and cerebrovascular expert. The label determined by the scheme for the user is more detailed, and not only floats on the large range of the surface, but also is a specific small label falling on the large range. For example, in the embodiment, in the determined reading tags of the users, the small tags of specific expert celebrities interested in the reading tags are further refined from the big tags of the celebrities; in addition, the preferred text mode of the user is determined. Preferably, frequency statistics is performed on specific small labels, so that the attention of the user to each type of small labels (keywords) can be obtained. Therefore, more refined matching can be performed according to the label of the user, and topics which are more concerned and more interesting to the user can be matched, so that recommendation can be performed. Certainly, since each topic generated by each neural network model is related to the selected topic keyword, if the matching degree of the selected topic keyword and the user is not high, the topic matched with the user label can be searched from the past historical topic posts, and then recommendation is performed.
In another embodiment of the topic recommendation method based on the user tag, as shown in fig. 2, on the basis of the above embodiment, the reading tag of the user includes: reading keywords and frequency thereof, and text mode preference; the topic recommendation method of the embodiment specifically includes:
s201, selecting topic keywords from a keyword tag library in a cloud; different keywords and frequency numbers are stored in the keyword tag library in a classified manner;
s202, inputting the topic keywords into trained texts in different modes to generate a neural network model, and obtaining topics in different modes of the topic keywords;
s203, acquiring the reading label of the user,
s204, acquiring a keyword with the highest frequency number in the reading label of the user as a target keyword;
s205, searching topics matched with the target keywords;
s206, when the topics matched with the target keywords are found, selecting the topics matched with the text mode preference in the reading label of the user from the found topics as target topics;
s207, recommending the target topic to the user.
In this embodiment, after the reading tag of the user is obtained, the keyword with the highest frequency may be locked according to the reading tag of the user, and the keyword may be used as the target keyword. Since the highest frequency indicates that the user is most concerned, the user is most likely to participate in the discussion if the recommended topic is related to the target keyword. In addition, after the topics matched with the target keywords are found, text modes (tone or writing) of the topics are different, and preferences of different users to the text modes are different.
In any of the above embodiments, how the text-generating neural network model of each different mode is trained specifically includes:
s301, crawling hot topics with reading amount exceeding a preset threshold from the Internet;
s302, classifying the collected hot topics according to a text mode;
s303, extracting keywords in the hot topics, labeling and obtaining training samples;
s304, inputting the training samples into different neural network models according to the types of the text modes for iterative training, and obtaining the text generation neural network models of different modes.
Specifically, at the beginning of training, a large number of hot topics need to be crawled from the internet, and the fact that the participation degree of a user who selects the hot topics is high means that the user is generally favored by the user but favored by the user. Therefore, the embodiment crawls a large number of hot topics and has very representative and learning significance. After a large amount of hot topic materials are obtained, the hot topic materials need to be processed, specifically, the hot topics need to be classified according to a text mode, that is, according to an expression method (a tone or a writing method) of the hot topic materials, and in addition, keywords of the hot topic materials need to be extracted from the hot topic materials and labeled. And finally, respectively inputting the neural network models of the corresponding categories according to the categories for training. For example, the hot topic training sample of the emotional pattern is input into the neural network model of the emotional pattern for training, and after training, the topic related to the input keyword can be output through the neural network model of the emotional pattern, and the expression pattern of the topic is the emotional pattern.
Another embodiment of the topic recommendation method based on the user tag according to the present application is shown in fig. 3, and includes:
s401, selecting topic keywords from a keyword tag library in a cloud; different keywords and frequency numbers are stored in the keyword tag library in a classified manner;
s402, inputting the topic keywords into trained texts in different modes to generate a neural network model, and obtaining topics in different modes of the topic keywords;
s403, acquiring a reading label of the user;
s404, whether a target topic matched with the reading label of the user is found or not is judged according to the reading label of the user; if yes, go to step S408; otherwise, go to step S405;
s405, determining a target keyword and a text mode according to the reading label of the user;
s406, selecting a text generation neural network model of a corresponding mode as a target text generation neural network model according to the determined text mode;
s407, inputting the determined target keywords into the target text to generate a neural network model, and generating a target topic in a corresponding mode;
s408, recommending the target topic to the user.
In this embodiment, when a topic matched with the reading tag of the user is not found (a topic that can be matched with the reading tag of the user is referred to as a target topic), a target keyword and a favorite text mode that the user is interested in are determined according to the reading tag of the user; and selecting a text of the text mode to generate a neural network model, inputting the determined target keyword into the model, further outputting the topic which is interested by the user and is about the target keyword, wherein the text mode of the topic is favored by the user, and finally recommending the generated target topic to the user.
Of course, when the matched target topic is not found from the existing topics according to the reading tags of the user, in addition to the implementation method of the above embodiment, other implementation methods may be adopted, specifically, for example, when the topic matched with the reading tags of the user is not found, all topics in the website are sorted according to the reading amount or the participation amount, so as to obtain a plurality of hot topics; the number of hot topics are then recommended to the user. The scheme is very suitable for the situation that the topic matched with the reading label of the user is not found or the reading label of the user cannot be obtained, and as all topics on the website are sequenced according to the participation situation or the reading quantity, the topic which is actively participated in or interested by a plurality of users can be known, and the topic recommended to the user has higher possibility of participating in the hot topic compared with other common topics (non-hot topics).
According to the other embodiment of the user tag-based topic recommendation method, topic citation is added on the basis of any one of the embodiments. That is to say, in the embodiment of the application, in addition to generating the target topic for the selected target keyword, the method may further obtain an extension keyword of the target keyword according to the keyword extension map stored in the keyword tag library, and further obtain a corresponding topic related to the extension keyword through the text generation neural network model according to the extension keyword. Since the extension keyword is an extension of and related to the target keyword, if the user is interested in the topic of the target keyword, the user is also likely to be interested in the extension topic of the target keyword. Therefore, when recommending topics to a user, in addition to recommending topics of direct target keywords, topics of extension keywords related to the target keywords can be recommended, so that the user can participate in a plurality of different interesting topics.
Specifically, the present embodiment has several implementation forms as follows:
(1) if the topic matched with the reading label of the user is found according to the reading label of the user, extracting a target keyword of the topic according to the found topic; then obtaining an extension keyword of the target keyword according to the keyword extension map; after the extension keyword is obtained, whether a topic related to the extension keyword exists is searched according to the extension keyword. In this case, the following cases are divided:
1, finding topics related to the extension key words in existing topics, and further judging whether a text mode of the found topics of the extension key words is a favorite text mode in a user reading label; if yes, the topic of the searched extension keyword and the topic which is searched for and matched according to the reading label of the user are taken as the target topic and recommended to the user.
Further, if the text mode of the topic of the reiteration keyword is not the text mode which is favored by the user in the reading label, the text of the text mode (determined according to the reading label of the user) corresponding to the reiteration keyword can be input to generate a neural network model, so that the topic which is favored by the user and is related to the reiteration keyword is generated, and finally, the generated topic and the topic which is searched and matched according to the reading label of the user are taken together as the target topic and recommended to the user.
Of course, if the text mode of the topic of the extension keyword is not the text mode which is favored by the user in the reading label, the topic recommendation of the extension keyword can be given up, and only the topic which is searched for and matched according to the reading label of the user is taken as the target topic and recommended to the user.
If the topic related to the extension keyword is not found from the existing topics, the extension keyword can be input into the text of the corresponding text mode (determined according to the reading label of the user) to generate a neural network model, then the topic about the extension keyword in the text mode preferred by the user is generated, and finally the generated topic and the topic which is found and matched according to the reading label of the user are taken together as the target topic to be recommended to the user.
(2) If the topic matched with the reading label of the user is not found according to the reading label of the user, acquiring a target keyword interested by the user and a favorite text mode according to the reading label of the user; selecting a text of a corresponding mode according to the favorite text mode to generate a neural network model; acquiring an extension keyword of the target keyword according to the keyword extension map; finally, inputting the target key words and the extension key words into texts in selected corresponding modes to generate a neural network model, so as to obtain topics of the target key words and topics of the extension key words; and the generated text patterns of the topics are also liked by the user (according with the reading label of the user), and finally the generated topics are recommended to the user as the target topics together.
In the embodiment, the extension keywords of the target keywords are obtained based on the keyword extension map, so that more recommended topics can be obtained, and the extension keywords are associated with the target keywords and belong to extension expansion of the target keywords, so that the related topics are easy to attract attention and interest of users, and the topic recommendation range is widened on the premise of ensuring the recommendation accuracy.
In addition, the keyword extension map in the embodiment can display extension expansion entities of each target keyword; for example, if the target keyword is "caesarean delivery", the keyword may be "post-partum repair", "scar-removing medicine", or the like.
In another embodiment of the topic recommendation system based on the user tag, on the basis of any one of the above method embodiments, a similar topic processing step is added, and specifically, the topic recommendation method of this embodiment includes the following steps:
s501, selecting topic keywords from a keyword tag library in the cloud; different keywords and frequency numbers are stored in the keyword tag library in a classified manner;
s502, inputting the topic keywords into trained texts in different modes to generate a neural network model, and obtaining topics in different modes of the topic keywords;
s503, searching topics related to the keywords from the published topics as alternative topics;
s504, screening out topics with similarity reaching a preset threshold value with the alternative topics from the topics acquired through the neural network model, and deleting the topics;
s505, the reading label of the user is obtained,
s506, searching topics matched with the reading labels of the users according to the reading labels of the users, and recommending the topics to the users.
In the embodiment, after the neural network model is generated through texts in different modes to obtain topics in different modes of topic keywords, similarity matching is carried out on the generated topics and the existing topics; if the topics with larger similarity degree (the similarity degree reaches a preset threshold value) exist, deleting the topics with larger similarity degree obtained through the neural network model, and only keeping the topics with the similarity degree not higher than the preset threshold value. Specifically, for example, topic keywords a are screened out from a keyword tag library, and then the topic keywords a are respectively input into texts with different modes to generate neural network models X1, X2 and X3; thereby obtaining the topic about the topic keyword A of the corresponding mode: a1, a2, a 3; if the topic keyword A is published in the topics published before, the topic keyword A also comprises the following topics: a '1, a' 2, a '3, a' 4, a '5, a' 6; comparing the similarity of the topics a1, a2 and a3 with the topics a ' 1, a ' 2, a ' 3, a ' 4, a ' 5 and a ' 6, respectively, and if the similarity of the topic a1 and the topic a ' 1 reaches a preset threshold value after comparison, deleting the topic a1, and reserving the topic related to the topic keyword a: a2, a3, a ' 2, a ' 3, a ' 4, a ' 5, a ' 6; finally, according to the reading label of the user, the topic matched with the reading label of the user is searched from the topics a2, a3, a ' 2, a ' 3, a ' 4, a ' 5 and a ' 6, and the topic is recommended to the user.
Based on the same technical concept, the application also discloses a topic recommendation system based on the user label, and the system can recommend topics to the user by adopting the method implemented by any topic recommendation method. Specifically, as shown in fig. 4, an embodiment of the user tag-based topic recommendation system of the present application includes:
a keyword tag library 10 for storing different keywords and frequency counts in a classified manner; the keyword tag library 10 is established in the cloud; specifically, the keyword tag library 10 stored in the cloud stores different keywords and the frequency count of the keywords according to categories. The frequency is the number of statistics for the keyword.
The keyword selection module 20 is configured to select topic keywords from the cloud keyword tag library 10; specifically, the topic keywords may be selected according to the frequency of each keyword stored therein, and a keyword with a high frequency indicates that the attention degree of the keyword in the network community is relatively high, so that the topic about the keyword is most likely to be a hot topic. Of course, the topic keywords may be selected according to the frequency of the keywords, or the topic keywords may be selected according to the frequency increase rate (the frequency increase rate of the keywords in the preset time period may be calculated according to the frequency of the keywords at the historical time point stored in the keyword tag library 10).
The topic generation module 30 is configured to input the topic keywords into trained texts in different modes to generate a neural network model, so as to obtain topics in different modes of the topic keywords; the mode is a discourse mode or a writing method, and the effects obtained by using different writing methods are different for the same keyword and the same thing. The schema types can be defined on demand. Inputting the same key words, the different patterns of text generation neural networks generate different patterns (different writing methods or different tones) of topics.
A tag obtaining module 40, configured to obtain a reading tag of a user; for example, the reading tag of the user may be determined according to the historical reading information of the user, the registration information of the user, and the like.
And the topic recommending module 50 is configured to search a topic matched with the reading tag of the user according to the reading tag of the user, and recommend the topic to the user. Specifically, after the reading label of the user is determined, the reading label can be searched and matched from various topics according to the reading label, and finally, the matched topic is recommended to the user to guide the user to participate in topic discussion. Matching topics, namely, matching is only carried out from the topics generated by the text generation neural network models; another implementation manner is that, in addition to the topics generated by the neural network model generated by each text according to the selected topic keywords in the embodiment, the topics may also be historical topics published before. The former has small matching workload; the latter match is more likely to succeed. Of course, matching may be performed from the generated topics first, and if the generated topics do not match, matching may be performed from the historical topics.
In this embodiment, the keyword selection module 20 selects topic keywords from the keyword tag library 10, the topic generation module 30 generates different types of topics by using different patterns of text generation neural network models, then the tag acquisition module 40 acquires the reading tags of the user, the topic recommendation module 50 performs matching according to the reading tags of the user, and selects topics in a pen mode that the user may be interested in and likes to recommend, so that the possibility of the user participating in the topic discussion is greatly improved, and further the activity of the network community is improved.
In another embodiment of the system of the present application, on the basis of the above system embodiment, as shown in fig. 5, the tag obtaining module 40 includes:
a historical data acquisition submodule 41, configured to acquire search and read historical data information of the user;
an extraction statistics submodule 42, configured to extract keywords from the historical data information, and count frequency numbers of the keywords;
a data analysis sub-module 43, configured to obtain a text mode preference of the user according to the historical data information; and determining the reading label of the user according to the counted keywords, the frequency and the preference of the text mode.
Preferably, the reading tag of the user includes: reading keywords and frequency thereof, and text mode preference; the topic recommendation module 50 specifically includes:
a target obtaining submodule 51, configured to obtain a keyword with a highest frequency number in the reading tag of the user as a target keyword;
the matching search submodule 52 is used for searching topics matched with the target keywords;
a topic selection submodule 53, configured to, when topics matched with the target keyword are found, select, from the found topics, a topic matched with a text mode preference in the reading tag of the user as a target topic;
and the information recommending submodule 54 is used for recommending the target topic to the user.
In this embodiment, the topic recommendation module 50 in the above system embodiment is refined, specifically, after the target obtaining sub-module 51 obtains the reading tag of the user, the keyword with the highest frequency is locked according to the reading tag of the user, and the keyword can be used as the target keyword. Since the highest frequency indicates that the user is most concerned, the user is most likely to participate in the discussion if the recommended topic is related to the target keyword. In addition, when the matching search sub-module 52 searches topics matched with the target keyword, since text patterns (tone or writing methods) of the topics are also different, and preferences of different users to the text patterns are also different, in this embodiment, in addition to considering that the users may be interested in the topics of the target keyword, preferences of the users to the text patterns are further considered, the topic selection sub-module 53 further selects the topics of the text patterns liked by the users as the target topics from the searched topics related to the target keyword, and finally, recommends through the information recommendation sub-module 54. Therefore, not only the user is interested in the content of the target keyword, but also the user likes the stroke writing method, so that the user is more favored, the user is more easily triggered, and the user is more willing to participate in the discussion of the topic.
The above embodiment illustrates the following topic selection and recommendation operation when the topic matching the target keyword is found, and how should the topic matching the target keyword be operated if the topic matching the target keyword is not found? The following embodiment describes in detail the processing steps when no topic matching the target keyword is found. Specifically, on the basis of the previous embodiment, the topic recommendation system of this embodiment, the topic recommendation module 50 further includes: model selection submodule 55, wherein:
the target obtaining sub-module 51 is further configured to determine a target keyword and a text mode according to the reading tag of the user when the matching search sub-module 52 does not search the topic matched with the target keyword;
the model selection submodule 55 is configured to select, according to the determined text mode, a text generation neural network model in a corresponding mode as a target text generation neural network model; inputting the determined target keywords into the target text through the topic generation module 30 to generate a neural network model, and generating a target topic in a corresponding mode;
the information recommending submodule 54 is further configured to recommend the target topic to the user.
In this embodiment, when a topic matched with the reading tag of the user is not found (a topic that can be matched with the reading tag of the user is referred to as a target topic), a target keyword and a favorite text mode that the user is interested in are determined according to the reading tag of the user; and selecting a text of the text mode to generate a neural network model, inputting the determined target keyword into the model, further outputting the topic which is interested by the user and is about the target keyword, wherein the text mode of the topic is favored by the user, and finally recommending the generated target topic to the user.
Of course, when the matching target topic is not found from the existing topics according to the reading label of the user, in addition to the implementation method in the above embodiment, other implementation methods may be adopted, for example, another embodiment of the system of the present application is shown in fig. 6, and on the basis of any of the above embodiments, the topic recommendation system based on the user label further includes:
the topic sorting module 70 is configured to, when a topic matched with the reading tag of the user is not found, sort all topics in the website according to the reading amount or the participation amount, and obtain a plurality of hot topics;
the topic recommendation module 70 is further configured to recommend the plurality of hot topics to the user.
In the embodiment, when the topic matched with the reading label of the user is not found, all topics in the website are sequenced according to the reading amount or the participation amount, and a plurality of hot topics are obtained; the number of hot topics are then recommended to the user. The scheme is very suitable for the situation that the topic matched with the reading label of the user is not found or the reading label of the user cannot be obtained, and as all topics on the website are sequenced according to the participation situation or the reading quantity, the topic which is actively participated in or interested by a plurality of users can be known, and the topic recommended to the user has higher possibility of participating in the hot topic compared with other common topics (non-hot topics).
In another embodiment of the system of the present application, on the basis of any of the above embodiments, the user tag-based topic recommendation system further includes:
a model training module 60, configured to train texts in different modes to generate a neural network model; the model training module 60 specifically includes:
the information crawling submodule 61 is used for crawling hot topics with reading amount exceeding a preset threshold value from the internet;
the information processing submodule 62 is used for classifying the collected hot topics according to the text mode; extracting keywords in the hot topics for labeling to obtain training samples;
and the iterative training submodule 63 is configured to input the training samples into different neural network models according to the types of the text modes to perform iterative training, so as to obtain text generation neural network models with different modes.
Specifically, in this embodiment, in addition to the technical features of any one of the above system embodiments, a model training module in the system is mainly described, and the model training module is used to train texts in different modes to generate a neural network model.
In a final embodiment of the system of the present application, as shown in the figure, on the basis of any one of the above system embodiments, the keyword tag library 10 further stores a keyword extension map; the topic recommendation module 50 further includes: an association lookup sub-module 56; wherein:
the correlation search sub-module 56 is configured to search an extension keyword of the target keyword according to the keyword extension map;
the matching search sub-module 52 is further configured to search topics matching the target keywords or the extension keywords;
the topic selection sub-module 53 is further configured to select, from the searched topics, a topic that matches the text pattern preference in the user's reading tag as a target topic when the topic that matches the target keyword or the extension keyword is found; and finally recommending the target topic to the user through the information recommending submodule 54.
In the embodiment of the application, in addition to generating the target topic for the selected target keyword, the reiteration keyword of the target keyword can be obtained according to the keyword reiteration map stored in the keyword tag library 10, and then the corresponding topic related to the reiteration keyword is obtained through the text generation neural network model according to the reiteration keyword. Since the extension keyword is an extension of and related to the target keyword, if the user is interested in the topic of the target keyword, the user is also likely to be interested in the extension topic of the target keyword. Therefore, when recommending topics to a user, in addition to recommending topics of direct target keywords, topics of extension keywords related to the target keywords can be recommended, so that the user can participate in a plurality of different interesting topics.
The system embodiment of the present application corresponds to the method embodiment, and the technical details of the method embodiment of the present application are also applicable to the method embodiment of the present application, and are not described again to reduce repetition.
While the preferred embodiments of the present application 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 alterations and modifications as fall within the scope of the application.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present application without departing from the spirit and scope of the application. Thus, if such modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is intended to include such modifications and variations as well.
Claims (15)
1. A topic recommendation method based on user tags is characterized by comprising the following steps:
selecting topic keywords from a keyword tag library in the cloud; different keywords and frequency numbers are stored in the keyword tag library in a classified manner;
inputting the topic keywords into trained texts in different modes to generate a neural network model, and obtaining topics in different modes of the topic keywords;
the reading label of the user is obtained,
and searching topics matched with the reading labels of the users according to the reading labels of the users, and recommending the topics to the users.
2. The user tag-based topic recommendation method according to claim 1, further comprising:
acquiring historical data information of searching and reading of the user;
extracting keywords from the historical data information, and counting the frequency of each keyword;
acquiring the text mode preference of the user according to the historical data information;
and determining the reading label of the user according to the counted keywords, the counted frequency and the preference of the text mode.
3. The user tag-based topic recommendation method according to claim 1, wherein the reading tag of the user comprises: reading keywords and frequency thereof, and text mode preference; the searching for the topic matched with the reading label of the user according to the reading label of the user and recommending the topic to the user specifically comprises:
acquiring a keyword with the highest frequency number in the reading label of the user as a target keyword;
searching topics matched with the target keywords;
when the topics matched with the target keywords are found, selecting the topics matched with the text mode preference in the reading label of the user as target topics from the found topics;
recommending the target topic to the user.
4. The user tag-based topic recommendation method according to claim 1, further comprising:
training texts in different modes to generate a neural network model; the method specifically comprises the following steps:
crawling hot topics with reading quantity exceeding a preset threshold from the internet;
classifying the collected hot topics according to a text mode;
extracting keywords in the hot topics, and labeling to obtain training samples;
inputting the training samples into different neural network models according to the types of the text modes for iterative training, and obtaining text generation neural network models of different modes.
5. The user tag-based topic recommendation method according to claim 1, further comprising:
when the topic matched with the reading label of the user is not found, determining a target keyword and a text mode according to the reading label of the user;
selecting a text generation neural network model of a corresponding mode as a target text generation neural network model according to the determined text mode;
inputting the determined target keywords into the target text to generate a neural network model, and generating a target topic in a corresponding mode;
recommending the target topic to the user.
6. The user tag-based topic recommendation method according to claim 1, further comprising:
when the topic matched with the reading label of the user is not found, sequencing all topics in the website according to the reading amount or participation amount to obtain a plurality of hot topics;
recommending the number of hot topics to the user.
7. The user tag-based topic recommendation method according to claim 3, wherein the keyword tag library further stores a keyword extension map; after the keyword with the highest frequency number in the reading label of the user is obtained as the target keyword, the method comprises the following steps:
searching for an extension keyword of the target keyword according to the keyword extension map;
searching topics matched with the target keywords or the extension keywords;
when topics matched with the target keywords or the extension keywords are found, selecting topics matched with the text mode preference in the reading label of the user from the found topics as target topics;
recommending the target topic to the user.
8. The method for recommending topics based on user tags as claimed in claim 1, further comprising after obtaining topics of different patterns of the topic keywords:
searching topics relevant to the keywords from published topics as alternative topics;
and screening out topics with similarity reaching a preset threshold value with the alternative topics from the topics acquired through the neural network model, and deleting the topics.
9. A user tag based topic recommendation system, comprising:
the keyword tag library is used for storing different keywords and frequency numbers in a classified manner; the keyword tag library is established in the cloud;
the keyword selection module is used for selecting topic keywords from a keyword tag library in the cloud;
the topic generation module is used for inputting the topic keywords into trained texts in different modes to generate a neural network model so as to obtain topics in different modes of the topic keywords;
the tag acquisition module is used for acquiring a reading tag of a user;
and the topic recommendation module is used for searching topics matched with the reading labels of the users according to the reading labels of the users and recommending the topics to the users.
10. The system of claim 9, wherein the tag obtaining module comprises:
the historical data acquisition submodule is used for acquiring searching and reading historical data information of the user;
the extraction and statistics submodule is used for extracting keywords from the historical data information and counting the frequency of each keyword;
the data analysis submodule is used for acquiring the text mode preference of the user according to the historical data information; and determining the reading label of the user according to the counted keywords, the frequency and the preference of the text mode.
11. The user tag-based topic recommendation system of claim 9, wherein the user's reading tag comprises: reading keywords and frequency thereof, and text mode preference; the topic recommendation module specifically comprises:
the target acquisition submodule is used for acquiring a keyword with the highest frequency number in the reading label of the user as a target keyword;
the matching search sub-module is used for searching topics matched with the target keywords;
the topic selection submodule is used for selecting topics matched with the text mode preference in the reading label of the user as target topics from the searched topics when the topics matched with the target keywords are searched;
and the information recommendation submodule is used for recommending the target topic to the user.
12. The user tag-based topic recommendation system of claim 9, further comprising:
the model training module is used for training texts in different modes to generate a neural network model; the model training module specifically comprises:
the information crawling sub-module is used for crawling hot topics with reading quantity exceeding a preset threshold value from the internet;
the information processing submodule is used for classifying the collected hot topics according to the text mode; extracting keywords in the hot topics for labeling to obtain training samples;
and the iterative training submodule is used for inputting the training samples into different neural network models according to the types of the text modes for iterative training to obtain the text generation neural network models of different modes.
13. The user tag-based topic recommendation system of claim 11, wherein the topic recommendation module further comprises: a model selection submodule, wherein:
the target obtaining sub-module is further used for determining a target keyword and a text mode according to the reading label of the user when the matching searching sub-module does not find the topic matched with the target keyword;
the model selection submodule is used for selecting a text generation neural network model of a corresponding mode as a target text generation neural network model according to the determined text mode; inputting the determined target keywords into the target text through the topic generation module to generate a neural network model, and generating a target topic in a corresponding mode;
the information recommending submodule is also used for recommending the target topic to the user.
14. The user tag-based topic recommendation system of claim 9, further comprising:
the topic sequencing module is used for sequencing all topics in the website according to the reading amount or participation amount when the topics matched with the reading tags of the users are not found, and obtaining a plurality of hot topics;
the topic recommending module is further used for recommending the hot topics to the user.
15. The system for recommending topics based on user tags according to any one of claims 11-14, wherein the keyword tag library further stores a keyword extension map; the topic recommendation module further comprises: an association search submodule; wherein:
the correlation searching sub-module is used for searching an extension keyword of the target keyword according to the keyword extension map;
the matching search sub-module is also used for searching topics matched with the target keywords or the extension keywords;
the topic selection sub-module is further configured to select, from the searched topics, a topic that matches the text mode preference in the reading tag of the user as a target topic when the topics that match the target keyword or the extension keyword are found.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010787292.3A CN111914079A (en) | 2020-08-07 | 2020-08-07 | Topic recommendation method and system based on user tags |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010787292.3A CN111914079A (en) | 2020-08-07 | 2020-08-07 | Topic recommendation method and system based on user tags |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111914079A true CN111914079A (en) | 2020-11-10 |
Family
ID=73282947
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010787292.3A Pending CN111914079A (en) | 2020-08-07 | 2020-08-07 | Topic recommendation method and system based on user tags |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111914079A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113434633A (en) * | 2021-06-28 | 2021-09-24 | 平安科技（深圳）有限公司 | Social topic recommendation method, device, equipment and storage medium based on head portrait |
| WO2023045345A1 (en) * | 2021-09-22 | 2023-03-30 | 华为云计算技术有限公司 | Topic mining method and related apparatus |
| CN116760882A (en) * | 2023-08-18 | 2023-09-15 | 广州朗歌信息技术有限公司 | Multimedia information distribution supervision system and method based on Internet of things |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8032506B1 (en) * | 2000-08-25 | 2011-10-04 | Andrej Gregov | User-directed product recommendations |
| CN103886090B (en) * | 2014-03-31 | 2018-01-02 | 北京搜狗科技发展有限公司 | Content recommendation method and device based on user preferences |
| CN108255999A (en) * | 2017-12-29 | 2018-07-06 | 北京奇虎科技有限公司 | Content recommendation method and device |
| CN108846097A (en) * | 2018-06-15 | 2018-11-20 | 北京搜狐新媒体信息技术有限公司 | The interest tags representation method of user, article recommended method and device, equipment |
| CN109190046A (en) * | 2018-09-18 | 2019-01-11 | 北京点网聚科技有限公司 | Content recommendation method, device and content recommendation service device |
| CN110413875A (en) * | 2019-06-26 | 2019-11-05 | 腾讯科技（深圳）有限公司 | A kind of method and relevant apparatus of text information push |
-
2020
- 2020-08-07 CN CN202010787292.3A patent/CN111914079A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8032506B1 (en) * | 2000-08-25 | 2011-10-04 | Andrej Gregov | User-directed product recommendations |
| CN103886090B (en) * | 2014-03-31 | 2018-01-02 | 北京搜狗科技发展有限公司 | Content recommendation method and device based on user preferences |
| CN108255999A (en) * | 2017-12-29 | 2018-07-06 | 北京奇虎科技有限公司 | Content recommendation method and device |
| CN108846097A (en) * | 2018-06-15 | 2018-11-20 | 北京搜狐新媒体信息技术有限公司 | The interest tags representation method of user, article recommended method and device, equipment |
| CN109190046A (en) * | 2018-09-18 | 2019-01-11 | 北京点网聚科技有限公司 | Content recommendation method, device and content recommendation service device |
| CN110413875A (en) * | 2019-06-26 | 2019-11-05 | 腾讯科技（深圳）有限公司 | A kind of method and relevant apparatus of text information push |
Non-Patent Citations (2)
| Title |
|---|
| 唐铮: "《微传播、观念、话语和模式》", 30 April 2016, 中国广播影视出版社 * |
| 谭贤: "《新媒体运营实战 从入门到精通》", 31 January 2019, 中国铁道出版社 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113434633A (en) * | 2021-06-28 | 2021-09-24 | 平安科技（深圳）有限公司 | Social topic recommendation method, device, equipment and storage medium based on head portrait |
| WO2023045345A1 (en) * | 2021-09-22 | 2023-03-30 | 华为云计算技术有限公司 | Topic mining method and related apparatus |
| CN116760882A (en) * | 2023-08-18 | 2023-09-15 | 广州朗歌信息技术有限公司 | Multimedia information distribution supervision system and method based on Internet of things |
| CN116760882B (en) * | 2023-08-18 | 2023-10-31 | 广州朗歌信息技术有限公司 | Multimedia information distribution supervision system and method based on Internet of things |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110059271B (en) | Searching method and device applying tag knowledge network | |
| CN111191122A (en) | Learning resource recommendation system based on user portrait | |
| CN106709040B (en) | Application search method and server | |
| CN103714130B (en) | Video recommendation system and method | |
| JP4637969B1 (en) | Properly understand the intent of web pages and user preferences, and recommend the best information in real time | |
| CN102483745B (en) | Co-selected image classification | |
| CN111914079A (en) | Topic recommendation method and system based on user tags | |
| CN104484431B (en) | A kind of multi-source Personalize News webpage recommending method based on domain body | |
| CN104866554B (en) | A kind of individuation search method and system based on socialization mark | |
| CN112434151A (en) | Patent recommendation method and device, computer equipment and storage medium | |
| CN107885793A (en) | A kind of hot microblog topic analyzing and predicting method and system | |
| CN111177559B (en) | Text travel service recommendation method and device, electronic equipment and storage medium | |
| CN111914172B (en) | Medical information recommendation method and system based on user tags | |
| CN111191111B (en) | Content recommendation method, device and storage medium | |
| Okazaki et al. | How to mine brand Tweets: Procedural guidelines and pretest | |
| CN106789559A (en) | Information processing method, device and system for wechat public platform | |
| CN108090228B (en) | Method and device for interaction through cultural cloud platform | |
| KR101559719B1 (en) | Auto-learning system and method for derive effective marketing | |
| CN111191112A (en) | Electronic reading data processing method, device and storage medium | |
| CN111061954B (en) | Search result sorting method and device and storage medium | |
| US20160048575A1 (en) | System and method for topics extraction and filtering | |
| JP7166116B2 (en) | Information processing device, information processing method, and program | |
| CN110609950A (en) | Public opinion system search word recommendation method and system | |
| CN116882414B (en) | Automatic comment generation method and related device based on large-scale language model | |
| CN110175289A (en) | Mixed recommendation method based on cosine similarity collaborative filtering |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20201110 |
|
| RJ01 | Rejection of invention patent application after publication |