CN111737558A - Information recommendation method and device and computer readable storage medium - Google Patents
Information recommendation method and device and computer readable storage medium Download PDFInfo
- Publication number
- CN111737558A CN111737558A CN202010436120.1A CN202010436120A CN111737558A CN 111737558 A CN111737558 A CN 111737558A CN 202010436120 A CN202010436120 A CN 202010436120A CN 111737558 A CN111737558 A CN 111737558A
- Authority
- CN
- China
- Prior art keywords
- information
- user
- similarity
- recommendation list
- piece
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 54
- 238000003860 storage Methods 0.000 title claims abstract description 13
- 239000011159 matrix material Substances 0.000 claims abstract description 127
- 230000006399 behavior Effects 0.000 claims description 91
- 238000004364 calculation method Methods 0.000 claims description 43
- 238000001914 filtration Methods 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 11
- 230000011218 segmentation Effects 0.000 claims description 9
- 238000001816 cooling Methods 0.000 claims description 7
- 230000004927 fusion Effects 0.000 claims description 4
- 238000012163 sequencing technique Methods 0.000 claims description 4
- 230000000694 effects Effects 0.000 abstract description 8
- 238000010586 diagram Methods 0.000 description 10
- 230000008569 process Effects 0.000 description 9
- 238000004590 computer program Methods 0.000 description 7
- 238000012986 modification Methods 0.000 description 5
- 230000004048 modification Effects 0.000 description 5
- 238000004422 calculation algorithm Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 238000010276 construction Methods 0.000 description 2
- 239000013598 vector Substances 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002457 bidirectional effect Effects 0.000 description 1
- 210000001072 colon Anatomy 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—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
- G06F16/9532—Query formulation
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/237—Lexical tools
- G06F40/242—Dictionaries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/284—Lexical analysis, e.g. tokenisation or collocates
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses an information recommendation method, an information recommendation device and a computer readable storage medium, which relate to the technical field of Internet, and the method comprises the following steps: generating a first information recommendation list corresponding to the current user based on information related to historical behavior data of the current user and a pre-constructed first similarity matrix, wherein each similarity in the first similarity matrix represents the content similarity between every two pieces of information; generating a second information recommendation list corresponding to the current user based on a pre-constructed user information preference matrix and a second similarity matrix, wherein each similarity in the second similarity matrix is obtained based on the number of common preference users owned by every two pieces of information; and fusing the first information recommendation list and the second information recommendation list to obtain a final information recommendation list to be recommended to the current user. The embodiment of the invention can improve the information recommendation effect.
Description
Technical Field
The present invention relates to the field of internet technologies, and in particular, to an information recommendation method, an information recommendation apparatus, and a computer-readable storage medium.
Background
With the rapid development of big data technology and artificial intelligence technology, finding out topics in which users are interested from massive information data and recommending related information to users become a research hotspot. However, due to the large number of articles and the large number of users, it is difficult to recommend corresponding information articles to users individually.
In the prior art, when information is recommended to a user in an individualized manner, information recommendation can be performed according to past behavior habits of the user, but only information articles related to preferences existing in historical behaviors can be recommended to the user in this manner, interest exploration requirements of the user cannot be met well, the recommendation effect is poor, namely the information articles recommended to the user cannot be guaranteed to meet the interests of the user, and meanwhile the diversity of the recommended information articles can be improved.
Disclosure of Invention
In order to solve the problems in the related art, embodiments of the present invention provide an information recommendation method, an information recommendation apparatus, and a computer-readable storage medium, which can ensure that an information article recommended to a user meets the interest of the user, and can improve the diversity of the recommended information article, thereby improving the information recommendation effect.
The embodiment of the invention provides the following specific technical scheme:
in a first aspect, an information recommendation method is provided, the method including:
generating a first information recommendation list corresponding to a current user based on information related to historical behavior data of the current user and a pre-constructed first similarity matrix, wherein each similarity in the first similarity matrix represents the content similarity degree between every two pieces of information;
generating a second information recommendation list corresponding to the current user based on a pre-constructed user information preference matrix and a second similarity matrix, wherein each similarity in the second similarity matrix is obtained based on the number of common preference users owned by every two pieces of information;
and fusing the first information recommendation list and the second information recommendation list to obtain a final information recommendation list to be recommended to the current user.
Further, the first similarity matrix is constructed by:
extracting keywords and corresponding keyword word frequencies of each piece of information in the information set;
sorting the keywords of each piece of information according to the value of the keyword frequency to obtain a keyword sorting list corresponding to each piece of information;
according to the keyword ranking list corresponding to each piece of information, calculating the similarity among all pieces of information according to a first calculation formula, and constructing an initial information similarity matrix;
and sequencing the similarity of each row in the initial information similarity matrix according to the size, and determining the sequenced initial information similarity matrix as the first similarity matrix.
Preferably, the first calculation formula is:
Scoreij＝∑win×wjmTin＝Tjm；
wherein, ScoreijIs the similarity between information i and information j, winThe word frequency, w, of the nth keyword of the information ijmThe word frequency, T, of the mth keyword of information jinIs the nth keyword, T, of the information ijmIs the m-th keyword of the information j.
Further, the extracting of the keywords and the corresponding keyword frequencies of each information in the information set includes:
performing word segmentation processing on the information and filtering stop words to obtain a plurality of candidate words of the information aiming at each information in the information set; and
and matching the candidate words with a preset keyword word bank, determining the candidate words which are successfully matched as the keywords of the information, and counting the keyword word frequency corresponding to each keyword of the information.
Preferably, the method further comprises:
acquiring the average word frequency of each piece of information according to the word frequency of the keyword corresponding to each keyword of each piece of information;
and filtering the keywords of each piece of information based on the average word frequency number of each piece of information.
Further, the user information preference matrix is constructed by the following method:
for the historical behavior data of each user in the user set, the following operations are performed:
extracting a plurality of preference behavior data of the user corresponding to each information in the information set from the historical behavior data of the user;
according to the preset weight corresponding to each preference behavior data, performing weighted calculation on each preference behavior data corresponding to each piece of information by the user to obtain the interest degree value of each piece of information by the user;
and constructing the user information preference matrix based on the interest degree value of each user to each piece of information.
Further, the second similarity matrix is constructed by:
according to the historical behavior data of the user corresponding to each piece of information in the information set, counting the number of the preferred users of each piece of information:
and calculating the similarity between every two pieces of information according to a second calculation formula according to the number of the preferred users of each piece of information and the number of the preferred users common to every two pieces of information so as to construct the second similarity matrix.
Preferably, the second calculation formula is:
wherein, wijIs the similarity between information i and information j, NiNumber of users, N, of preference information ijNumber of users, N, of preference information ji∩NjThe number of users who prefer both information i and information j.
Further, the generating a second information recommendation list corresponding to the current user based on the pre-constructed user information preference matrix and the second similarity matrix includes:
determining a plurality of candidate information of the current user without historical behaviors from an information set;
for each candidate information, determining similar information with the similarity degree of the candidate information ranked at the top K bits from the second similarity matrix, and determining common information between the similar information and the related information;
for each candidate information, calculating the interest degree value of the current user for each candidate information according to a third calculation formula according to the similarity between the candidate information and each common information and the interest degree value of the current user for each common information in the user information preference matrix;
and sorting the interest degree values of the current user to each candidate information according to the size to generate the second information recommendation list.
Preferably, the third calculation formula is:
Scorestj＝∑i∈N(t)∩S(j,k)wjiγti；
wherein, the ScorestjIs the interest degree value of user t to information j, N (t) is the information set with positive user behavior, S (j, k) is the information set with similarity degree of information j ranked k first, wjiIs the similarity of information i and information j, gammatiIs the interest level of the user t in the information i.
Further, the fusing the first information recommendation list and the second information recommendation list to obtain a final information recommendation list to be recommended to the current user includes:
extracting the common information in the first information recommendation list and the second information recommendation list;
and sorting the common information, and sequentially adding partial residual information in the first information recommendation list and/or the second information recommendation list to the sorted common information to obtain the final information recommendation list.
Further, the method further comprises:
when the current user is a new user, acquiring a current day heat value of each piece of information calculated based on Newton's cooling law and user historical behavior data corresponding to each piece of information in the information set;
and sequencing all the information in the information set according to the current heavyweight value of each piece of information to generate an information recommendation list and recommend the information recommendation list to the current user.
In a second aspect, an information recommendation apparatus is provided, the apparatus comprising:
the first generation module is used for generating a first information recommendation list corresponding to the current user based on information related to historical behavior data of the current user and a pre-constructed first similarity matrix, wherein each similarity in the first similarity matrix represents the content similarity between every two pieces of information;
the second generation module is used for generating a second information recommendation list corresponding to the current user based on a pre-constructed user information preference matrix and a second similarity matrix, wherein each similarity in the second similarity matrix is obtained based on the number of common preference users owned by every two pieces of information;
the information fusion module is used for fusing the first information recommendation list and the second information recommendation list to obtain a final information recommendation list;
and the information recommending module is used for recommending the final information recommending list to the current user.
In a third aspect, an information recommendation apparatus is provided, the apparatus comprising:
one or more processors;
a memory;
the program stored in the memory, when executed by the one or more processors, causes the processors to perform the steps of the information recommendation method of any one of the first aspects above.
In a fourth aspect, there is provided a computer-readable storage medium storing a program which, when executed by a processor, causes the processor to perform the steps of the information recommendation method according to any one of the first aspect.
The embodiment of the invention provides an information recommendation method, an information recommendation device and a computer readable storage medium, wherein each similarity in a first similarity matrix represents the content similarity between every two pieces of information, so that a first information recommendation list generated based on information related to historical behavior data of a user and a pre-constructed first similarity matrix is recommended to the current user, and information articles recommended to the user can be ensured to meet the user interest; moreover, each similarity in the second similarity matrix is obtained based on the number of the common preferred users owned by each two pieces of information, namely, when more users prefer two pieces of information at the same time, the similarity between the two pieces of information is higher, if the current user prefers one of the two pieces of information, the current user can be inferred to have a preference demand on the other piece of information in the two pieces of information, therefore, by recommending the second information recommendation list generated based on the user information preference matrix and the second similarity matrix to the current user, the information articles recommended to the user can be ensured to meet the interests of the user, meanwhile, the diversity of the recommended information articles is also ensured, and the cross-domain information recommendation is also facilitated to be realized, so the information recommendation lists obtained for the first information recommendation list and the second information recommendation list are recommended to the user, the information recommendation result is more diversified and has higher precision, so that the information recommendation effect can be improved.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only one embodiment of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIG. 1 is a flowchart illustrating an information recommendation method according to an embodiment of the present invention;
FIG. 2 is a schematic flow chart illustrating the construction of the first similarity matrix in step 102 shown in FIG. 1;
FIG. 3 is a schematic flow chart of step 201 shown in FIG. 2;
FIG. 4 is a flowchart illustrating the step 102 of constructing the user information preference matrix shown in FIG. 1;
FIG. 5 is a schematic flow chart illustrating the construction of the second similarity matrix in step 102 shown in FIG. 1;
FIG. 6 is a schematic flow chart of step 102 shown in FIG. 1;
FIG. 7 is a schematic flow chart of step 103 shown in FIG. 1;
fig. 8 is a schematic structural diagram of an information recommendation apparatus according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It is to be understood that, unless the context clearly requires otherwise, throughout the description and the claims, the words "comprise", "comprising", and the like are to be construed in an inclusive sense as opposed to an exclusive or exhaustive sense; that is, what is meant is "including, but not limited to".
Furthermore, in the description of the present invention, it is to be understood that the terms "first," "second," and the like are used for descriptive purposes only and are not to be construed as indicating or implying relative importance. In addition, in the description of the present invention, "a plurality" means two or more unless otherwise specified.
Fig. 1 is a flowchart illustrating an information recommendation method according to an embodiment of the present invention. The information recommendation method is exemplified by being applied to an information recommendation device, and the information recommendation device can be configured in any computer equipment, so that the computer equipment can execute the information recommendation method.
Referring to fig. 1, an information recommendation method according to an embodiment of the present invention may include the following steps:
101, generating a first information recommendation list corresponding to the current user based on information related to historical behavior data of the current user and a pre-constructed first similarity matrix, wherein each similarity in the first similarity matrix represents a content similarity degree between every two pieces of information.
Here, the current user may be a user who refreshes the information flow through the client at the current time. The historical behavior data of the user is information of behavior operation performed by the user, and the information may be information of behavior operation performed by the user at the previous moment before the current moment, or information of behavior operation performed by the user at other moments before the current moment, where the behavior operation includes, but is not limited to, various operation behaviors of browsing, clicking, commenting, collecting, forwarding, sharing, and the like of the user on the information. The information may be financial information such as stock information and fund information, scientific information, entertainment information, and other types of information.
The process of constructing the first similarity matrix may be to extract keywords and keyword word frequencies of each piece of information, sort the keywords according to the keyword word frequencies, calculate the similarity between each two pieces of information based on a keyword sorted list between each two pieces of information in combination with the keyword word frequencies, and construct the first similarity matrix. In addition, the method may also be implemented in other manners, and the embodiment does not limit the specific process.
In this embodiment, information related to historical behavior data of the current user may be determined, and according to the similarity in the first similarity matrix, a plurality of pieces of information to be recommended that are most similar to the information related to the historical behavior data are determined, and a first information recommendation list is generated, wherein if the similarity between the information to be recommended and the information related to the historical behavior data of the current user is greater, the position of the information to be recommended in the first information recommendation list is further forward.
And 102, generating a second information recommendation list corresponding to the current user based on a pre-constructed user information preference matrix and a second similarity matrix, wherein each similarity in the second similarity matrix is obtained based on the number of the preference users shared by every two pieces of information.
Wherein, each interest degree value in the user information preference matrix represents the interest degree of the user for the information.
The process of constructing the user information preference matrix may be: the method comprises the steps of extracting a plurality of preference behavior data of each user to each piece of information from historical behavior data of each user in a user set within a preset time period, carrying out weighted calculation on the plurality of preference behavior data to obtain an interest degree value of the user to the information, and further constructing and obtaining a user information preference matrix, wherein the preference behavior data can comprise one or more of browsing times, clicking times, comment times, collection times, forwarding times and sharing times. In addition, other methods may be used to construct the user information preference matrix, and the embodiment does not limit the specific process.
The process of constructing the second similarity matrix may be: when the similarity between two pieces of information is calculated, the similarity between the two pieces of information is calculated by using the number of the respective preferred users of the two pieces of information and the number of the common preferred users of the two pieces of information, so as to construct and obtain a second similarity matrix. Here, when a user simultaneously prefers two different pieces of information, the user is a common preferred user owned by the two pieces of information.
In this embodiment, information that the current user is interested in may be determined from the user information preference matrix, and according to the similarity in the second similarity matrix, a plurality of pieces of information to be recommended that are most similar to the information that the current user is interested in are determined, and a second information recommendation list is generated, where the greater the similarity between the information to be recommended and the information that the current user is interested in, the more forward the position of the candidate information in the second information recommendation list is.
103, the first information recommendation list and the second information recommendation list are fused to obtain a final information recommendation list to be recommended to the current user.
In this embodiment, a plurality of pieces of information to be recommended included in the first information recommendation list and included in the second information recommendation list may be extracted first, and the plurality of pieces of information to be recommended are sorted to generate a final information recommendation list to be recommended to the user.
The embodiment of the invention provides an information recommendation method, because each similarity in a first similarity matrix represents the content similarity between every two pieces of information, a first information recommendation list generated based on information related to historical behavior data of a user and a first similarity matrix constructed in advance is recommended to the current user, and information articles recommended to the user can be ensured to meet the user interest; moreover, each similarity in the second similarity matrix is obtained based on the number of the common preferred users owned by each two pieces of information, namely, when more users prefer two pieces of information at the same time, the similarity between the two pieces of information is higher, if the current user prefers one of the two pieces of information, the current user can be inferred to have a preference demand on the other piece of information in the two pieces of information, therefore, by recommending the second information recommendation list generated based on the user information preference matrix and the second similarity matrix to the current user, the information articles recommended to the user can be ensured to meet the interests of the user, meanwhile, the diversity of the recommended information articles is also ensured, and the cross-domain information recommendation is also facilitated to be realized, so the information recommendation lists obtained for the first information recommendation list and the second information recommendation list are recommended to the user, the information recommendation result is more diversified and has higher precision, so that the information recommendation effect can be improved.
In addition, considering that there is no historical behavior data of the current user, that is, when the system is in cold boot, the current user is a new user who logs in to view information for the first time, and the above embodiment is not suitable for recommending information to the new user, so in order to avoid the problem of cold boot, the method provided by the embodiment of the present invention may further include a step of recommending information based on a recommendation algorithm for heat prediction, in addition to the above steps.
In one embodiment of the present invention, the method may further comprise:
when the current user is a new user, acquiring a current day heat value of each piece of information calculated based on Newton's cooling law and user historical behavior data corresponding to each piece of information in the information set;
and sorting all the information in the information set according to the current heavyweight value of each piece of information to generate an information recommendation list and recommending the information recommendation list to the current user.
Specifically, when the historical behavior data of the current user on the information does not exist in the buried point database, the current user is determined to be a new user.
Wherein, the calculation process of the current day heat value of each information in the information set may be: various types of user behavior data such as browsing, praise and collection of all information in the information set in a past time period (for example, the last 7 days) are obtained, different weight values are distributed according to the types of the user behavior data, for example, the weights of the collection, praise and browsing from high to low are distributed, and the heat data of each information is calculated.
Since the heat of an information is gradually reduced in the life cycle except for the presentation state, similar to the newton's law of cooling in nature, the heat of the day can be predicted by using the newton's law of cooling, and the heat value of the information of the day can be calculated by using the following calculation formula:
wherein, ScoreiThe heat of the day i, ratio, interval, and Score are the cooling coefficient, interval, and the information heat of the day obtained from the previous D days. Preferably, the ratio is 0.02 and the interval is 24.
Therefore, the daily popularity value of each piece of information is obtained through calculation, all the information is sorted, N pieces of information with the largest daily popularity prediction are found, when a new user logs in a client to browse the information, the information with the largest daily popularity prediction is displayed for the user, the cold start problem can be avoided, and information recommendation can be carried out on the user without historical preference behavior data.
In an embodiment of the present invention, as shown in fig. 2, the first similarity matrix in step 102 is constructed by the following steps:
201, extracting keywords and corresponding keyword frequencies of each information in the information set.
202, sorting the keywords of each piece of information according to the value of the keyword frequency to obtain a keyword sorting list corresponding to each piece of information.
203, according to the keyword ranking list corresponding to each information, calculating the similarity between all the information according to the first calculation formula, and constructing an initial information similarity matrix.
Preferably, the first calculation formula is:
Scoreij＝∑win×wjmTin＝Tjm；
wherein, ScoreijIs the similarity between information i and information j, winThe word frequency, w, of the nth keyword of the information ijmThe word frequency, T, of the mth keyword of information jinIs the nth keyword, T, of the information ijmIs the m-th keyword of the information j.
It is understood that the similarity between the information may also be calculated by other calculation methods, which is not limited in this embodiment.
In the embodiment of the invention, in the process of calculating the information similarity, compared with the conventional recommendation algorithm, the obtained long sentence is converted into the vector, and the distance between word vectors is calculated, the keyword of the information and the word frequency thereof are extracted, and the keywords of each information are sorted from large to small according to the value of the word frequency to obtain the keyword sorted list, so that the similarity between the information can be calculated directly based on the keyword sorted list of the information, the data calculation amount is reduced, the simplicity, the high efficiency and the high precision are realized, and the calculation effect of the similarity can be improved.
204, sorting the similarity of each row in the initial information similarity matrix according to size, and determining the sorted initial information similarity matrix as a first similarity matrix.
In this embodiment, each similarity in the first similarity matrix represents a content similarity between every two pieces of information, and the similarity in each row in the matrix is obtained by sorting according to the similarity, so that when it is determined that the information related to the historical behavior data of the current user is the information i (for example, the current user browses, forwards, approves or collects the information i in the previous moment of the current moment), the first information recommendation list may be generated by determining a matrix row corresponding to the information i in the first similarity matrix, and sequentially determining, from the head of the matrix row, a plurality of pieces of information to be recommended with the highest similarity to the information i in the matrix row.
It is understood that when new information is added to the information set, steps 201 to 204 are performed for the new information to update the first similarity matrix.
In an embodiment of the present invention, as shown in fig. 3, the step 201 of extracting the keywords and the corresponding keyword frequencies of each information in the information set may include the steps of:
301, for each information in the information set, performing word segmentation processing on the information and filtering out stop words to obtain multiple candidate words of the information.
Specifically, the Chinese word segmentation algorithm based on character string matching performs word segmentation processing on the text part of each information article. For example, a bidirectional maximum matching algorithm based on string matching, that is, forward maximum matching and reverse maximum matching are simultaneously calculated, so as to obtain word segmentation results of each piece of information, and a constructed stop word lexicon is used to filter stop words of the word segmentation results of each piece of information, so as to obtain a plurality of candidate words of each piece of information.
The stop word lexicon includes words that are frequently used but have no meaning and common punctuation marks, such as commas, periods, colons, numbers, "my", "we", "also", "of", etc.
302, matching the candidate words with a preset keyword lexicon, determining the candidate words successfully matched as the keywords of the information, and counting the keyword word frequency corresponding to each keyword of the information.
Specifically, a preset keyword lexicon can be constructed based on the actual information recommendation purpose, the number of the keyword lexicons can be one or more, when the number of the keyword lexicons is multiple, multiple candidate words of one piece of information need to be matched with each keyword lexicon respectively, and the candidate words which are successfully matched with each keyword lexicon are used as the keywords of the piece of information. For example, if the information recommendation purpose is to recommend fund and stock information articles, a word stock of listed companies can be formed by extracting the company name, stock code, high-management name, important shareholder name and other words of the listed companies in China; a fund product word library is formed by extracting domestic fund data comprising fund names, fund short names, fund codes, fund managers and other words.
After determining the keywords of each piece of information, the word frequency of the keywords corresponding to each keyword of each piece of information can be counted, so that each keyword of each piece of information and the corresponding word frequency thereof are obtained and stored in a dictionary form.
In the embodiment, the information content is subjected to keyword extraction by combining the keyword lexicon, a limited amount of keywords are used for representing the information content, the data volume is reduced, and the subsequent calculation of the similarity between information is facilitated.
In addition, in order to avoid that the keyword in the partial information generates excessive noise due to an error of the word segmentation model and affects the accuracy of the similarity calculation of the subsequent information, the implementation process of step 201 may further include steps 303 to 304 after determining the keyword of each information and the corresponding keyword word frequency.
303, obtaining the average word frequency of each information according to the word frequency of the keyword corresponding to each keyword of each information.
Wherein, wikIs the word frequency corresponding to the k-th keyword of the information i, n is the number of keywords of the information i,
the calculated average word frequency is obtained.
304, filtering the keywords of each information based on the average word frequency of each information.
Specifically, when filtering the keywords of the information i, the word frequency ratio of the keywords in the information i can be obtained
Filtering all small keywords and keeping the word frequency of the keywords not less than
The keywords of each piece of information after filtering processing are sorted according to the value of the keyword frequency, and a keyword sorting list of each piece of information is obtained.
In the embodiment, the keywords of the information are subjected to low-frequency filtering and other processing according to the word frequency of the keywords, the word frequency sorting is utilized to obtain the keyword list, and then the similarity between the information is calculated, so that the similarity calculation effect can be improved through the rough sorting of the characteristic engineering.
In an embodiment of the present invention, as shown in fig. 4, the user information preference matrix in step 102 is constructed by the following steps:
401, for the historical behavior data of each user in the user set, extracting a plurality of preference behavior data of the user corresponding to each information in the information set from the historical behavior data of the user.
Specifically, for each user in the user set, historical behavior data of the user on each piece of information in the information set within a preset historical time period is acquired from a database in which buried point data is stored, preference behavior data corresponding to the information is extracted from the historical behavior data, and the preference behavior data refers to data capable of representing that the user has preference behaviors on the information, such as behavior data of browsing, collecting, forwarding, sharing, agreeing on and the like of the user on the information.
402, according to the preset weight corresponding to each preference behavior data, performing weighted calculation on each preference behavior data corresponding to each information of the user to obtain the interest degree value of each information of the user.
Wherein different weights may be assigned to different types of preference behavior in advance.
Specifically, value domain distribution of each type of preference behavior data is counted, abnormal values in the preference behavior data of the user are removed, weighting calculation is carried out on each preference behavior data corresponding to each information according to preset weight corresponding to each preference behavior data, and the interest degree value of the user on each information is obtained.
Illustratively, three kinds of preference behavior data are taken, for example: browsing, praise and collection, the corresponding weight can be set as: the collection behavior has the highest corresponding weight, the following is praise, and the lowest is browsing.
The interest degree value of the user to the information can be calculated by the following calculation formula:
Scoreij＝Favoriteij*α+Likeij*β+Browseij*γ；
wherein i is the ith subscriber, j is the information j, FavoriteijFor i useNumber of times of collection of information j by user, LikeijBrowse for i number of user's approval of information jiji the browsing times of the user to the information j, preferably, α, β, γ are respectively and correspondingly taken as 10, 5, 1.
It should be noted that, although three preference behaviors of browsing, favoring and collecting are taken as examples to calculate the interest level value of the user in the information in the embodiment, it is understood that the method provided by the embodiment of the present invention is not limited thereto, and those skilled in the art may also calculate the interest level value of the user in the information by using more types of preference behaviors.
403, constructing a user information preference matrix based on the interest degree value of each user for each information.
In this embodiment, the interest level value of each user for each information is obtained through calculation, and finally a user preference information matrix of user-information is formed, where the rows of the matrix are "user number", the behaviors of the matrix are "information number", and the values in the matrix are "interest level value of user for information".
In an embodiment of the present invention, as shown in fig. 5, the second similarity matrix in step 102 is constructed by the following steps:
501, counting the number of the preferred users of each information according to the historical behavior data of the user corresponding to each information in the information set.
Specifically, the preferred behavior data of different users corresponding to each information is extracted from the historical behavior data corresponding to each information in the preset historical time period, and the number of the preferred users of each information is counted according to the preferred behavior data of different users corresponding to each information.
The preferred behavior data refers to data that can represent a user's preferred behavior for information, such as behavior data of browsing, collecting, forwarding, sharing, and agreeing to the information.
502, according to the number of the preferred users of each information and the number of the preferred users common to each two information, calculating the similarity between each two information according to a second calculation formula to construct a second similarity matrix.
Preferably, the second calculation formula is:
wherein, wijIs the similarity between information i and information j, NiNumber of users, N, of preference information ijNumber of users, N, of preference information ji∩NjThe number of users who prefer both information i and information j.
It can be known from the second calculation formula that, when the number of the common preferred users owned by the information i and the information j is larger, i.e. the number of the users preferring the information i and the information j is larger, the similarity w between the information i and the information j is largerijThe higher.
It is understood that when new information is added to the information set, steps 501 to 502 are performed to update the second similarity matrix for the new information.
In an embodiment of the present invention, as shown in fig. 6, the step 102 of generating a second information recommendation list corresponding to the current user based on the pre-constructed user information preference matrix and the second similarity matrix may include:
a plurality of candidate information items of which the current user has no historical behavior are determined 601 from the information set.
For each candidate information, the similarity information with the candidate information ranked K bits in the similarity degree from the second similarity matrix is determined, and common information is determined from the similarity information and the related information.
In this embodiment, the matrix rows in which the candidate information are respectively located may be determined in the second similarity matrix, and for each candidate information, k pieces of similar information with the similarity between the candidate information and the k bits in the top of the matrix rows that are located are determined, and information common to the k pieces of similar information and the information related to the current user pair is determined, where k is a positive integer not less than 1.
603, for each candidate information, calculating the interest level value of the current user for each candidate information according to a third calculation formula based on the similarity between the candidate information and each common information and the interest level value of the current user for each common information in the user information preference matrix.
Specifically, a user number and an information number corresponding to the current user are determined according to information related to historical behavior data of the current user, and an interest degree value of the current user on the related information is determined from a user information preference matrix according to the determined user number and the determined information number.
And for each candidate information, multiplying the similarity between the candidate information and each common information by the interest degree value of the current user to each common information, summing the obtained multiplication results, taking the summation result as the interest degree value of the current user to the candidate information, and so on to obtain the interest degree value of the current user to each candidate information.
Preferably, the third calculation formula is:
Scorestj＝∑i∈N(t)∩s(j,k)wjiγti；
wherein, the ScorestjIs the interest degree value of user t to information j, N (t) is the information set with positive user behavior, S (j, k) is the information set with similarity degree of information j ranked k first, wjiIs the similarity of information i and information j, gammatiIs the interest level of the user t in the information i. Here, the positive user behavior may specifically be browsing, collecting, forwarding, sharing, like behavior.
For example, assuming that k is 3, the information set is { a, B, C, D, E, F }, and if the information corresponding to the current user preference behavior is { a, B, C, D }, the candidate information is { E, F }. For the candidate information E, determining the information with the similarity of the information E in the first 3 bits from the second similarity matrix as information A, information F and information B in sequence; determining the common information between { A, B, C, D } and { A, F, B } as { A, B }; then, according to the user information preference matrix, obtaining the interest degree value of the current user for the information A and the interest degree value of the current user for the information B, determining the similarity between the candidate information E and the information A, B respectively according to the second similarity matrix, then calculating the interest degree value of the current user for the candidate information E according to a third calculation formula, and so on, calculating the interest degree value of the current user for the candidate information F.
604, the interest degree values of the current user for each candidate information are sorted according to size, and a second information recommendation list is generated.
In this embodiment, the interest level value of the current user for each candidate information is calculated based on the interest level value of the current user for the relevant information and the similarity between the relevant information and the candidate information determined from the second similarity matrix, since each similarity in the second similarity matrix is obtained based on the number of common preferred users owned by each two information, if the current user prefers one of the two similar information, it can be inferred that the current user has a preference requirement for the other of the two similar information (i.e. the candidate information), therefore, the second information recommendation list is generated and recommended to the user through the sequence of the interest degree values of the candidate information, the information in the second information recommendation list recommended to the user can be guaranteed to meet the interest exploration requirement of the user, and cross-domain information recommendation is facilitated.
In an embodiment of the present invention, as shown in fig. 7, the step 103 of fusing the first information recommendation list and the second information recommendation list to obtain a final information recommendation list to be recommended to the current user includes:
701, extracting the common information in the first information recommendation list and the second information recommendation list.
Here, when a certain information is included in both the first information recommendation list and the second information recommendation list, the information is common information between the first information recommendation list and the second information recommendation list.
702, sorting the common information, and sequentially adding part of the remaining information in the first information recommendation list and/or the second information recommendation list to the sorted common information to obtain a final information recommendation list.
Specifically, according to the preset information recommendation quantity, part of the remaining information in the first information recommendation list and/or the second information recommendation list is extracted and sequentially added to the sorted common information to obtain the final information recommendation list. Here, the preset information recommendation number may be set according to actual needs, for example, the number of information in the final information recommendation list is set to 10.
In the embodiment of the invention, the first information recommendation list and the second information recommendation list are fused to obtain the final information recommendation list to be recommended to the current user, so that not only can the information articles recommended to the user be ensured to meet the interest of the user, but also the diversity of the recommended information articles can be improved, and the information recommendation effect is improved.
Fig. 8 is a schematic structural diagram of an information recommendation device according to an embodiment of the present invention. Referring to fig. 8, an embodiment of the present invention further provides an information recommendation apparatus, including:
the first generating module 81 is configured to generate a first information recommendation list corresponding to the current user based on information related to historical behavior data of the current user and a pre-constructed first similarity matrix, where each similarity in the first similarity matrix represents a content similarity between every two pieces of information;
a second generating module 82, configured to generate a second information recommendation list corresponding to the current user based on a pre-constructed user information preference matrix and a second similarity matrix, where each similarity in the second similarity matrix is obtained based on the number of common preferred users owned by each two pieces of information;
the information fusion module 83 is configured to fuse the first information recommendation list and the second information recommendation list to obtain a final information recommendation list;
and the information recommending module 84 is used for recommending the final information recommending list to the current user.
In one embodiment of the invention, the apparatus further comprises a first building block for:
extracting keywords and corresponding keyword word frequencies of each piece of information in the information set;
sorting the keywords of each piece of information according to the value of the keyword frequency to obtain a keyword sorting list corresponding to each piece of information;
calculating the similarity among all the information according to a first calculation formula and constructing an initial information similarity matrix according to the keyword ranking list corresponding to each information;
sequencing the similarity of each row in the initial information similarity matrix according to the size, and determining the sequenced initial information similarity matrix as a first similarity matrix;
preferably, the first calculation formula is:
Scoreij＝∑win×wjmTin＝Tjm；
wherein, ScoreijIs the similarity between information i and information j, winThe word frequency, w, of the nth keyword of the information ijmThe word frequency, T, of the mth keyword of information jinIs the nth keyword, T, of the information ijmIs the m-th keyword of the information j.
In an embodiment of the invention, the first building block is specifically configured to:
obtaining keywords and keyword frequencies corresponding to each piece of information in the information set, including:
performing word segmentation processing on the information and filtering stop words aiming at each information in the information set to obtain a plurality of candidate words of the information; and
matching the candidate words with a preset keyword word bank, determining the candidate words which are successfully matched as the keywords of the information, and counting the keyword word frequency corresponding to each keyword of the information;
preferably, the first building block is further configured to:
acquiring the average word frequency of each piece of information according to the word frequency of the keyword corresponding to each keyword of each piece of information;
and filtering the keywords of each piece of information based on the average word frequency of each piece of information.
In one embodiment of the invention, the apparatus further comprises a second building block for:
for the historical behavior data of each user in the user set, the following operations are performed:
extracting a plurality of preference behavior data of the user corresponding to each information in the information set from the historical behavior data of the user;
according to the preset weight corresponding to each piece of preference behavior data, performing weighted calculation on each piece of preference behavior data corresponding to each piece of information by the user to obtain the interest degree value of each piece of information by the user;
and constructing a user information preference matrix based on the interest degree value of each user to each information.
In one embodiment of the invention, the apparatus further comprises a third building block for:
according to the historical behavior data of the user corresponding to each information in the information set, counting the number of the preferred users of each information:
calculating the similarity between every two pieces of information according to a second calculation formula according to the number of the preference users of each piece of information and the number of the preference users common to every two pieces of information so as to construct a second similarity matrix;
preferably, the second calculation formula is:
wherein, wijIs the similarity between information i and information j, NiNumber of users, N, of preference information ijNumber of users, N, of preference information ji∩NjThe number of users who prefer both information i and information j.
In an embodiment of the present invention, the second generating module is specifically configured to:
determining a plurality of candidate information of the current user without historical behaviors from an information set;
for each candidate information, determining similar information with the similarity degree of the candidate information ranked at the top K bits from the second similarity matrix, and determining common information between the similar information and the related information;
for each candidate information, calculating the interest degree value of the current user for each candidate information according to a third calculation formula according to the similarity between the candidate information and each common information and the interest degree value of the current user for each common information in the user information preference matrix;
sorting the interest degree values of the current user to each candidate information according to the size to generate a second information recommendation list;
preferably, the third calculation formula is:
Scorestj＝∑i∈N(t)∩S(j,k)wjiγti；
wherein, the ScorestjIs the interest degree value of user t to information j, N (t) is the information set with positive user behavior, S (j, k) is the information set with similarity degree of information j ranked k first, wjiIs the similarity of information i and information j, gammatiIs the interest level of the user t in the information i.
In an embodiment of the present invention, the information fusion module 83 is specifically configured to:
extracting the common information in the first information recommendation list and the second information recommendation list;
and sorting the common information, and sequentially adding part of the rest information in the first information recommendation list and/or the second information recommendation list to the sorted common information to obtain a final information recommendation list.
In one embodiment of the present invention, the apparatus further comprises a third generating module, configured to:
when the current user is a new user, acquiring a current day heat value of each piece of information calculated based on Newton's cooling law and user historical behavior data corresponding to each piece of information in the information set;
and sorting all the information in the information set according to the current heavyweight value of each information to generate an information recommendation list.
The information recommendation module 84 is further configured to:
and recommending the information recommendation list generated by the third generation module to the current user.
The embodiment of the invention provides an information recommendation device, belongs to the same inventive concept as the information recommendation method provided by the embodiment of the invention, can execute the information recommendation method provided by the embodiment of the invention, and has the corresponding functional modules and beneficial effects of executing the information recommendation method. For details of the information recommendation method provided in the embodiment of the present invention, reference may be made to the technical details not described in detail in the embodiment.
In addition, an embodiment of the present invention further provides an information recommendation apparatus, including:
one or more processors;
a memory;
a program stored in the memory, which when executed by the one or more processors, causes the processors to perform the steps of any of the information recommendation methods described in the above embodiments.
In addition, an embodiment of the present invention further provides a computer-readable storage medium, in which a program is stored, and when the program is executed by a processor, the processor is enabled to execute the steps of the information recommendation method in any of the above embodiments.
As will be appreciated by one of skill in the art, embodiments of the present invention may be provided as a method, apparatus, or computer program product. Accordingly, embodiments of the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, embodiments of the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
Embodiments of the present invention are described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the embodiments of the invention.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.
Claims (10)
1. An information recommendation method, the method comprising:
generating a first information recommendation list corresponding to a current user based on information related to historical behavior data of the current user and a pre-constructed first similarity matrix, wherein each similarity in the first similarity matrix represents the content similarity degree between every two pieces of information;
generating a second information recommendation list corresponding to the current user based on a pre-constructed user information preference matrix and a second similarity matrix, wherein each similarity in the second similarity matrix is obtained based on the number of common preference users owned by every two pieces of information;
and fusing the first information recommendation list and the second information recommendation list to obtain a final information recommendation list to be recommended to the current user.
2. The method of claim 1, wherein the first similarity matrix is constructed by:
extracting keywords and corresponding keyword word frequencies of each piece of information in the information set;
sorting the keywords of each piece of information according to the value of the keyword frequency to obtain a keyword sorting list corresponding to each piece of information;
according to the keyword ranking list corresponding to each piece of information, calculating the similarity among all pieces of information according to a first calculation formula, and constructing an initial information similarity matrix;
sorting each row of similarity in the initial information similarity matrix according to size, and determining the sorted initial information similarity matrix as the first similarity matrix;
preferably, the first calculation formula is:
Scoreij＝∑win×wjmTin＝Tjm；
wherein, ScoreijIs the similarity between information i and information j, winThe word frequency, w, of the nth keyword of the information ijmThe word frequency, T, of the mth keyword of information jinIs the nth keyword, T, of the information ijmIs the m-th keyword of the information j.
3. The method of claim 2, wherein extracting the keywords and corresponding keyword frequencies for each information in the information set comprises:
performing word segmentation processing on the information and filtering stop words to obtain a plurality of candidate words of the information aiming at each information in the information set; and
matching the candidate words with a preset keyword word bank, determining the candidate words which are successfully matched as the keywords of the information, and counting the keyword word frequency corresponding to each keyword of the information;
preferably, the method further comprises:
acquiring the average word frequency of each piece of information according to the word frequency of the keyword corresponding to each keyword of each piece of information;
and filtering the keywords of each piece of information based on the average word frequency number of each piece of information.
4. The method of claim 1, wherein the user information preference matrix is constructed by:
for the historical behavior data of each user in the user set, the following operations are performed:
extracting a plurality of preference behavior data of the user corresponding to each information in the information set from the historical behavior data of the user;
according to the preset weight corresponding to each preference behavior data, performing weighted calculation on each preference behavior data corresponding to each piece of information by the user to obtain the interest degree value of each piece of information by the user;
and constructing the user information preference matrix based on the interest degree value of each user to each piece of information.
5. The method of claim 1, wherein the second similarity matrix is constructed by:
according to the historical behavior data of the user corresponding to each piece of information in the information set, counting the number of the preferred users of each piece of information:
calculating the similarity between every two pieces of information according to a second calculation formula according to the number of preference users of each piece of information and the number of preference users common to every two pieces of information so as to construct a second similarity matrix;
preferably, the second calculation formula is:
wherein, wijIs the similarity between information i and information j, NiNumber of users, N, of preference information ijNumber of users, N, of preference information ji∩NjThe number of users who prefer both information i and information j.
6. The method of claim 1, wherein generating a second information recommendation list corresponding to the current user based on a pre-constructed user information preference matrix and a second similarity matrix comprises:
determining a plurality of candidate information of the current user without historical behaviors from an information set;
for each candidate information, determining similar information with the similarity degree of the candidate information ranked at the top K bits from the second similarity matrix, and determining common information between the similar information and the related information;
for each candidate information, calculating the interest degree value of the current user for each candidate information according to a third calculation formula according to the similarity between the candidate information and each common information and the interest degree value of the current user for each common information in the user information preference matrix;
sorting the interest degree values of the current user to each candidate information according to the size to generate a second information recommendation list;
preferably, the third calculation formula is:
Scorestj＝∑i∈N(t)∩S(j,k)wjiγti；
wherein, the ScorestjIs the interest degree value of user t to information j, N (t) is the information set with positive user behavior, S (j, k) is the information set with similarity degree of information j ranked k first, wjiIs the similarity of information i and information j, gammatiIs the interest level of the user t in the information i.
7. The method of claim 1, wherein the fusing the first information recommendation list and the second information recommendation list to obtain a final information recommendation list to recommend to the current user comprises:
extracting the common information in the first information recommendation list and the second information recommendation list;
and sorting the common information, and sequentially adding partial residual information in the first information recommendation list and/or the second information recommendation list to the sorted common information to obtain the final information recommendation list.
8. The method of any of claims 1 to 7, further comprising:
when the current user is a new user, acquiring a current day heat value of each piece of information calculated based on Newton's cooling law and user historical behavior data corresponding to each piece of information in the information set;
and sequencing all the information in the information set according to the current heavyweight value of each piece of information to generate an information recommendation list and recommend the information recommendation list to the current user.
9. An information recommendation apparatus, comprising:
the first generation module is used for generating a first information recommendation list corresponding to the current user based on information related to historical behavior data of the current user and a pre-constructed first similarity matrix, wherein each similarity in the first similarity matrix represents the content similarity between every two pieces of information;
the second generation module is used for generating a second information recommendation list corresponding to the current user based on a pre-constructed user information preference matrix and a second similarity matrix, wherein each similarity in the second similarity matrix is obtained based on the number of common preference users owned by every two pieces of information;
the information fusion module is used for fusing the first information recommendation list and the second information recommendation list to obtain a final information recommendation list;
and the information recommending module is used for recommending the final information recommending list to the current user.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium stores a program which, when executed by a processor, causes the processor to execute the steps of the information recommendation method according to any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010436120.1A CN111737558A (en) | 2020-05-21 | 2020-05-21 | Information recommendation method and device and computer readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010436120.1A CN111737558A (en) | 2020-05-21 | 2020-05-21 | Information recommendation method and device and computer readable storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111737558A true CN111737558A (en) | 2020-10-02 |
Family
ID=72647544
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010436120.1A Pending CN111737558A (en) | 2020-05-21 | 2020-05-21 | Information recommendation method and device and computer readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111737558A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112163158A (en) * | 2020-10-09 | 2021-01-01 | 贵州电网有限责任公司 | Search recommendation method based on search history and browsing footprint |
| CN113221006A (en) * | 2021-05-24 | 2021-08-06 | 北京搜狐新媒体信息技术有限公司 | Article recommendation method and device, electronic equipment and computer storage medium |
| CN113868528A (en) * | 2021-09-29 | 2021-12-31 | 平安银行股份有限公司 | Information recommendation method and device, electronic equipment and readable storage medium |
| CN114003826A (en) * | 2021-12-31 | 2022-02-01 | 思创数码科技股份有限公司 | Resource directory recommendation method and device, readable storage medium and electronic equipment |
| CN115563397A (en) * | 2022-12-06 | 2023-01-03 | 福建慧政通信息科技有限公司 | Electronic file recommendation method and terminal |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107766379A (en) * | 2016-08-22 | 2018-03-06 | 中国电信股份有限公司 | A kind of recommendation method and apparatus of Web content |
| CN108615177A (en) * | 2018-04-09 | 2018-10-02 | 武汉理工大学 | Electric terminal personalized recommendation method based on weighting extraction interest-degree |
| CN110309410A (en) * | 2018-03-15 | 2019-10-08 | ***通信集团有限公司 | A kind of information recommended method, platform and computer readable storage medium |
| CN110866180A (en) * | 2019-10-12 | 2020-03-06 | 平安国际智慧城市科技股份有限公司 | Resource recommendation method, server and storage medium |
-
2020
- 2020-05-21 CN CN202010436120.1A patent/CN111737558A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107766379A (en) * | 2016-08-22 | 2018-03-06 | 中国电信股份有限公司 | A kind of recommendation method and apparatus of Web content |
| CN110309410A (en) * | 2018-03-15 | 2019-10-08 | ***通信集团有限公司 | A kind of information recommended method, platform and computer readable storage medium |
| CN108615177A (en) * | 2018-04-09 | 2018-10-02 | 武汉理工大学 | Electric terminal personalized recommendation method based on weighting extraction interest-degree |
| CN110866180A (en) * | 2019-10-12 | 2020-03-06 | 平安国际智慧城市科技股份有限公司 | Resource recommendation method, server and storage medium |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112163158A (en) * | 2020-10-09 | 2021-01-01 | 贵州电网有限责任公司 | Search recommendation method based on search history and browsing footprint |
| CN113221006A (en) * | 2021-05-24 | 2021-08-06 | 北京搜狐新媒体信息技术有限公司 | Article recommendation method and device, electronic equipment and computer storage medium |
| CN113221006B (en) * | 2021-05-24 | 2023-12-19 | 北京搜狐新媒体信息技术有限公司 | Article recommendation method and device, electronic equipment and computer storage medium |
| CN113868528A (en) * | 2021-09-29 | 2021-12-31 | 平安银行股份有限公司 | Information recommendation method and device, electronic equipment and readable storage medium |
| CN114003826A (en) * | 2021-12-31 | 2022-02-01 | 思创数码科技股份有限公司 | Resource directory recommendation method and device, readable storage medium and electronic equipment |
| CN115563397A (en) * | 2022-12-06 | 2023-01-03 | 福建慧政通信息科技有限公司 | Electronic file recommendation method and terminal |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111737558A (en) | Information recommendation method and device and computer readable storage medium | |
| CN109690529B (en) | Compiling documents into a timeline by event | |
| JP5391634B2 (en) | Selecting tags for a document through paragraph analysis | |
| US20160092781A1 (en) | Similarity metric relativized to a user's preferences | |
| EP1995669A1 (en) | Ontology-content-based filtering method for personalized newspapers | |
| US20100185623A1 (en) | Topical ranking in information retrieval | |
| CN109902823B (en) | Model training method and device based on generation countermeasure network | |
| Nguyen et al. | Npe: neural personalized embedding for collaborative filtering | |
| CN111597446B (en) | Content pushing method and device based on artificial intelligence, server and storage medium | |
| CN111651678B (en) | Personalized recommendation method based on knowledge graph | |
| CN110866102A (en) | Search processing method | |
| Huang et al. | Leveraging the crowd to improve feature-sentiment analysis of user reviews | |
| Li et al. | A hybrid model for experts finding in community question answering | |
| CN112182145A (en) | Text similarity determination method, device, equipment and storage medium | |
| Saia et al. | Semantic coherence-based user profile modeling in the recommender systems context | |
| US7444313B2 (en) | Systems and methods for optimizing decision graph collaborative filtering | |
| CN114168790A (en) | Personalized video recommendation method and system based on automatic feature combination | |
| CN115827990B (en) | Searching method and device | |
| Li et al. | Exploit latent Dirichlet allocation for collaborative filtering | |
| Tejasree et al. | An improved differential bond energy algorithm with fuzzy merging method to improve the document clustering for information mining | |
| CN110806861B (en) | API recommendation method and terminal combining user feedback information | |
| Anarfi et al. | desc2tag: A reinforcement learning approach to mashup tag recommendation | |
| Ferdous et al. | Assortment of bangladeshi e-commerce site reviews using machine learning approaches | |
| Cai et al. | Marriage recommendation algorithm based on KD-KNN-LR model | |
| Ali et al. | Identifying and Profiling User Interest over time using Social Data |
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: 20201002 |
|
| RJ01 | Rejection of invention patent application after publication |