WO2019037258A1 - 信息推荐的装置、方法、系统及计算机可读存储介质 - Google Patents

信息推荐的装置、方法、系统及计算机可读存储介质 Download PDF

Info

Publication number
WO2019037258A1
WO2019037258A1 PCT/CN2017/108761 CN2017108761W WO2019037258A1 WO 2019037258 A1 WO2019037258 A1 WO 2019037258A1 CN 2017108761 W CN2017108761 W CN 2017108761W WO 2019037258 A1 WO2019037258 A1 WO 2019037258A1
Authority
WO
WIPO (PCT)
Prior art keywords
user
information
analyzed
target object
preset
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.)
Ceased
Application number
PCT/CN2017/108761
Other languages
English (en)
French (fr)
Inventor
王健宗
吴天博
黄章成
肖京
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ping An Technology Shenzhen Co Ltd
Original Assignee
Ping An Technology Shenzhen Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Ping An Technology Shenzhen Co Ltd filed Critical Ping An Technology Shenzhen Co Ltd
Publication of WO2019037258A1 publication Critical patent/WO2019037258A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • G06F16/3331—Query processing
    • G06F16/334—Query execution
    • G06F16/3344—Query execution using natural language analysis
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • G06F16/337—Profile generation, learning or modification
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
    • G06F16/374—Thesaurus
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/20—Natural language analysis
    • G06F40/279—Recognition of textual entities
    • G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking

Definitions

  • the present application relates to the field of communications technologies, and in particular, to an apparatus, method, system, and computer readable storage medium for information recommendation.
  • the user's interest points are mainly obtained by analyzing the user's historical behavior, and then various information is pushed according to the user's interest points.
  • the historical behavior of the user can be obtained through social media (such as Weibo, blog, forum, podcast, etc.) used by the user.
  • the user's positive preference is obtained by analyzing the historical behavior of the user, for example, user A is on Weibo. On the release of more travel and food texts and photos, you can analyze that user A's positive preference is to like travel and food.
  • the user's historical behavior is limited, and it is not accurate to recommend personalized information to the user based on the historical behavior of the user's Internet access; on the other hand, the user's interest is relatively stable, and the social background is in dynamic change (for example, current Hot news or time has occurred, and it is not possible to accurately recommend personalized information to users simply by analyzing the user's historical behavior.
  • the purpose of the present application is to provide an apparatus, method, system and computer readable storage medium for information recommendation, which are intended to accurately recommend personalized information to a user.
  • the present application provides an apparatus for information recommendation, the information recommendation apparatus comprising a memory and a processor connected to the memory, wherein the memory stores information recommended to be run on the processor.
  • the system when the information recommendation system is executed by the processor, implements the following steps:
  • the present application further provides a method for information recommendation, and the method for information recommendation includes:
  • the present application further provides a system for information recommendation, where the information recommendation system includes:
  • a first determining module configured to determine an approximate object corresponding to the target object according to a mapping relationship between the predetermined target object and the approximated object, and use the target object and the determined approximated object as the target object before recommending the target object to the user Object to be analyzed;
  • An analysis module configured to obtain, from a predetermined data source, evaluation data of the object to be analyzed by each user in a preset time, analyze the evaluation data of each object to be analyzed according to a predetermined analysis rule, and obtain each object to be analyzed Corresponding forward tag value;
  • a second determining module configured to determine, according to a predetermined recommendation algorithm, whether to recommend the target object to the user, if the forward label value is greater than or equal to a preset threshold
  • a recommendation module if yes, determining that the user is a related user of the target object, and recommending the target object to the user.
  • the application further provides a computer readable storage medium on which a system for information recommendation is stored, and when the system of information recommendation is executed by the processor, the steps are implemented:
  • the beneficial effects of the present application are: before recommending a target object to a user, the present application determines an approximate object corresponding to the target object according to a predetermined mapping relationship, and treats both the target object and the approximated object as The analysis object is obtained, and the evaluation data of each object to be analyzed by the user within a preset time is obtained from the predetermined data source, and the evaluation data of each object to be analyzed is analyzed according to a predetermined analysis rule, and the forward label corresponding to each object to be analyzed is obtained.
  • the present application is based on analyzing the evaluation data of the target object and the approximate object. The way to recommend the target object to the user can accurately recommend personalized information to the user.
  • FIG. 1 is a schematic diagram of an optional application environment of each embodiment of the present application.
  • FIG. 2 is a schematic diagram of a hardware architecture of an apparatus for information recommendation in FIG. 1;
  • FIG. 3 is a schematic structural diagram of a predetermined structure word segmentation tree
  • FIG. 4 is a schematic flowchart of an embodiment of a method for information recommendation according to the present application.
  • first, second and the like in the present application are for the purpose of description only, and are not to be construed as indicating or implying their relative importance or implicitly indicating the number of technical features indicated. .
  • features defining “first” and “second” may include at least one of the features, either explicitly or implicitly.
  • the technical solutions between the various embodiments may be combined with each other, but must be based on the realization of those skilled in the art, and when the combination of the technical solutions is contradictory or impossible to implement, it should be considered that the combination of the technical solutions does not exist. Nor is it within the scope of protection required by this application.
  • FIG. 1 is a schematic diagram of an optional application environment of each embodiment of the present application.
  • the present application is applicable to an application environment including, but not limited to, device 1, user terminal 2, and network 3 of information recommendation.
  • the device 1 for information recommendation communicates with one or more user terminals 2 via the network 3.
  • the user terminal 2 includes, but is not limited to, any electronic product that can interact with a user through a keyboard, a mouse, a remote controller, a touch panel, or a voice control device, for example, a personal computer, a tablet, or a smart phone.
  • PDA Personal Digital Assistant
  • game consoles Internet Protocol Television (IPTV)
  • IPTV Internet Protocol Television
  • smart wearable devices navigation devices, etc.
  • mobile devices such as digital TVs, desktop computers, Fixed terminal for notebooks, servers, etc.
  • the network 3 may be an intranet, an Internet, a Global System of Mobile communication (GSM), a Wideband Code Division Multiple Access (WCDMA), or a 4G. Network, 5G network Wireless or wired networks such as network, Bluetooth, Wi-Fi, etc.
  • GSM Global System of Mobile communication
  • WCDMA Wideband Code Division Multiple Access
  • 4G. Network 5G network Wireless or wired networks such as network, Bluetooth, Wi-Fi, etc.
  • the device 1 for information recommendation is a device capable of automatically performing numerical calculation and/or information processing in accordance with an instruction set or stored in advance.
  • the information recommendation device 1 may be a computer, a single network server, a server group composed of multiple network servers, or a cloud-based cloud composed of a large number of hosts or network servers, where cloud computing is a distributed computing one.
  • cloud computing is a distributed computing one.
  • kind a super virtual computer consisting of a group of loosely coupled computers.
  • the device 1 for information recommendation may include, but is not limited to, a memory 11 that can be communicably connected to each other through a system bus.
  • Figure 2 only shows the device 1 with information recommendations for components 11-13, but it should be understood that not all illustrated components may be implemented, and more or fewer components may be implemented instead. .
  • the memory 11 includes at least one type of readable storage medium including a flash memory, a hard disk, a multimedia card, a card type memory (eg, SD or DX memory, etc.), and a random access memory (RAM). , static random access memory (SRAM), read only memory (ROM), electrically erasable programmable read only memory (EEPROM), programmable read only memory (PROM), magnetic memory, magnetic disk, optical disk, and the like.
  • the memory 11 may be an internal storage unit of the device 1 recommended by the information, such as a hard disk or memory of the device 1 recommended by the information.
  • the memory 11 may also be an external storage device of the information recommendation device 1, such as a plug-in hard disk equipped with the device 1 recommended by the information, and a smart memory card (Smart Media Card, SMC). ), Secure Digital (SD) card, Flash Card, etc.
  • the memory 11 can also include both the internal storage unit of the device 1 for information recommendation and its external storage device.
  • the memory 11 is generally used to store a system installed in the device 1 for information recommendation and various types of application software, such as program codes of the system for information recommendation. Further, the memory 11 can also be used to temporarily store various types of data that have been output or are to be output.
  • the processor 12 may be a Central Processing Unit (CPU), controller, microcontroller, microprocessor, or other data processing chip in some embodiments.
  • the processor 12 is typically used to control the overall operation of the device 1 for information recommendation, such as performing control and processing associated with communication with the user terminal 2.
  • the processor 12 is configured to run program code or process data stored in the memory 11, such as a system that runs the information recommendation.
  • the network interface 13 may comprise a wireless network interface or a wired network interface, which is typically used to establish a communication connection between the device 1 for information recommendation and other electronic devices.
  • the network interface 13 is mainly used to connect the device 1 recommended by the information to one or more of the user terminals 2 through the network 3 to establish a data transmission channel and a communication connection.
  • the information recommendation system is stored in the memory 11 and includes at least one computer readable instruction stored in the memory 11, the at least one computer readable instruction being executable by the processor 12 to implement information of embodiments of the present application. a preferred method; and the at least one computer readable instruction can be classified into different logic modules according to functions implemented by the various portions thereof, including the first The fixed module, the analysis module, the second determining module and the recommended module.
  • Step S1 Before recommending the target object to the user, determining an approximate object corresponding to the target object according to a mapping relationship between the predetermined target object and the approximated object, and using the target object and the determined approximated object as the object to be analyzed;
  • the target object may be location information (such as restaurant location, library location or attraction location, etc.), item information (such as clothing accessories, books or daily necessities, etc.), network information (such as news, websites or audio and video, etc.), and financial management. Information (such as stocks, securities, insurance products, etc.).
  • the approximate object is an object similar to the target object. For example, if the target object is a restaurant location, the corresponding approximate object may be a food street location, a restaurant location, etc., and the target object is a library location, and the corresponding approximate object may be a variety of bookstores. Location, bookstore location, museum location, etc.
  • the target object and its corresponding approximation object are mapped in advance, and the mapping relationship between the target object and the approximation object is stored, for example, in the form of a mapping table.
  • the approximate object that has a mapping relationship with the target object is obtained according to the target object, and both the target object and the approximated object are used as subsequent objects to be analyzed.
  • step S2 the evaluation data of each object to be analyzed by each user in a preset time is obtained from a predetermined data source, and the evaluation data of each object to be analyzed is analyzed according to a predetermined analysis rule, and corresponding to each object to be analyzed is obtained.
  • Forward tag value the evaluation data of each object to be analyzed by each user in a preset time is obtained from a predetermined data source, and the evaluation data of each object to be analyzed is analyzed according to a predetermined analysis rule, and corresponding to each object to be analyzed is obtained.
  • the predetermined data source is, for example, a catering platform, a microblog website, a forum website, a life insurance server, etc.
  • the evaluation data includes a comment and a rating
  • each user is obtained from the predetermined data source to the waiting time within a preset time.
  • the evaluation data made by the analysis object is obtained, for example, from the predetermined data source, the evaluation data of the object to be analyzed by each user within six months is obtained.
  • the forward label value is a value in which the evaluation data is a variable, and the larger the evaluation amount in which the evaluation is high in the evaluation data, the larger the forward label value.
  • the predetermined analysis rule preferably includes:
  • evaluation data of the object to be analyzed is a score, and the score of the object to be analyzed is less than the preset score, determining that the object to be analyzed is an invalid object, or if the score is greater than or equal to the preset score, determining that the evaluation data is a positive information class And calculating a forward label value corresponding to the evaluation data of the object to be analyzed according to a preset calculation rule;
  • the evaluation data of the object to be analyzed is a comment, parsing the core viewpoint information corresponding to the comment, and using the pre-trained classifier to identify the information pointing category corresponding to the core viewpoint information (where the information pointing category includes the positive information category, Negative information class), if the information pointing category is a negative information class, determining that the object to be analyzed is an invalid object, or if the information pointing category is a positive information class, calculating the object to be analyzed according to a preset calculation rule The forward tag value corresponding to the evaluation data.
  • Step S3 if the forward label value is greater than or equal to a preset threshold, determining whether to recommend the target object to the user according to a predetermined recommendation algorithm;
  • Step S4 if yes, determining that the user is a related user of the target object, and recommending the target object to the user.
  • a predetermined threshold is preset, when the forward label value of the object to be analyzed is greater than or equal to
  • the threshold is preset, it indicates that the object to be analyzed is better or is recognized by the user, and further considers whether to recommend the target object to the user according to a predetermined recommendation algorithm, and the forward label value of the object to be analyzed is smaller than the preset.
  • the threshold is exceeded, the target object is not considered to be recommended to the user.
  • the predetermined recommendation algorithm may be: calculating a recommended value based on the forward label value, and determining, by the recommended value, whether to recommend the target object to the user, specifically, when the recommended value is greater than or equal to the preset recommended value,
  • the target object is recommended to the user, for example, the recommendation information of the target object is sent to the user, and the target object is not recommended to the user when the recommended value is less than the preset recommended value.
  • the present embodiment determines an approximate object corresponding to the target object according to a predetermined mapping relationship before recommending the target object to the user, and takes both the target object and the approximated object as objects to be analyzed; from a predetermined data source.
  • the method specifically includes:
  • the recommended value of the target object is calculated based on the forward label value and the recommended value calculation formula constructed by the user data of the object to be analyzed, where the recommended value is greater than
  • the target object is recommended to the user, where the recommended value is calculated as:
  • u,t) ⁇ P(o
  • u, t) is a recommended value of the user u to the target object or the approximate object o in the background t
  • the ⁇ is a weight ( ⁇ is a constant)
  • u) is The forward label value of the user u to the target object or the approximate object o
  • ⁇ t ) is the probability that the target object or the approximate object o is selected by the user u
  • ⁇ t represents the topic distribution under the time series background t
  • C ⁇ c1, c2, . . .
  • cn ⁇ is the user-generated content of the user u in the context of the time t
  • n is the number of user-generated content
  • sim(.) is a user-generated content (eg, the user-published micro The rating of the blog or forum, etc.) is similar to the target object
  • sim(.) represents the similarity between a microblog and a target object
  • this embodiment adopts The value of the similarity is calculated using a TF-IDF (term frequency–inverse document frequency) model.
  • the foregoing preset calculation rule includes: calculating a first quantity of the evaluation data of the positive information category that the user makes for the object to be analyzed, and the a second quantity of evaluation data belonging to the positive information category made by the user for all objects to be analyzed; a third quantity of evaluation data belonging to the positive information category made by all users for the object to be analyzed, and all users for all The fourth quantity of the evaluation data belonging to the positive information class made by the object to be analyzed; dividing the first quantity by the second quantity to obtain the first forward label parameter, and substituting the third quantity and the fourth quantity into the preset formula
  • the B is the third quantity
  • the A is the fourth quantity
  • the first forward label parameter is multiplied by the second forward label parameter, and the evaluation data corresponding to the object to be analyzed is obtained.
  • Forward tag value For the second forward label parameter, the B is the third quantity, and the A is the fourth quantity; the first forward label parameter is multiplied by the second forward label parameter, and the evaluation data corresponding to the object to be analyzed
  • the classifier is a support vector machine classifier
  • the training process of the classifier includes:
  • a core sample of positive information eg, 10,000
  • positive information categories eg, a wide range of Ping An health insurance coverage, fast safe car brands
  • Information samples for example, Ping An Auto Insurance claims slow service, Ping An wealth management products have no promised high
  • the training ends, or, If the accuracy rate is less than the preset accuracy rate, the number of core view information samples of the positive information class and the core view information samples of the negative information class are increased, and the training is resumed until the accuracy of the trained classifier is greater than or equal to the preset accuracy rate. The training is over.
  • the preset accuracy eg. 0.98
  • the method specifically includes:
  • each word segment corresponding to the comment is constructed into a preset structure word segmentation tree, and each piece segment corresponding to the comment is constructed into a preset structure word segment tree to parse out the core point information corresponding to the comment.
  • the participles include words and words, for example, for the comment "** launched *** products are very good”, after the word segmentation results are "**”, "launch", "of”, "***", "very ", "good.”
  • the word segmentation processing of the comment comprises: splitting the comment by a preset type punctuation mark (for example: ",”, “.”, “!, “;”, etc.), for example, from the comment
  • a preset type punctuation mark for example: ",”, “.”, “!, “;”, etc.
  • the information between the starting position and the first preset type punctuation is a short sentence; if there is no preset type punctuation at the end of the comment, from the last preset type punctuation to the end of the comment
  • the message is a short sentence and is punctuated from the first preset type punctuation to the last last preset type punctuation
  • the information between the symbols, the information between each of the two preset types of punctuation marks is a short sentence; if the end position of the comment has a preset type punctuation, the first punctuation from the first preset type to the first from the last
  • the long word priority principle For each short sentence of the split, the long word priority principle is used to continue the word segmentation: for example, the long word priority principle means: for a phrase T1 that requires a word segment, starting with the first word A, from the pre-stored thesaurus Find the longest word X1 starting from A, then remove X1 from T1 and leave T2, and then use the same segmentation principle for T2.
  • the result after segmentation is "X1/X2/, ,,,,,,,,,,, for example, the result of the review "Ping An has introduced the product of Zunhong Life” is "Peace” / "Publish” / "Yes” / "Zhonghong Life” / "Product".
  • performing part-of-speech tagging on each participle corresponding to the comment includes:
  • mapping relationship between words and words in the universal word dictionary library and part of speech for example, in the universal word dictionary library, the part of speech corresponding to the playground is a noun
  • mapping relationship between the preset words and words and the part of speech for example, In the mapping relationship between the preset words and words and the part of speech, the part of speech corresponding to the playground is a common noun
  • the part of speech corresponding to each participle of each short sentence is determined, wherein the part of the word and the word are mapped to the part of speech.
  • the labeling priority is higher than the mapping between words and words in the universal word dictionary library and part of speech.
  • part of speech includes: real words (nouns, verbs, adjectives, quantifiers, pronouns, etc.), virtual words (adverbs, prepositions, conjunctions, auxiliary words, interjections, onomatopoeia, etc.).
  • each participle corresponding to the comment into a preset structure word segmentation tree includes:
  • the preset structure word segmentation tree includes a multi-level node, the first-level node is the comment itself, and the second-level node is a participle phrase obtained by the comment according to the order of the corresponding participle and part of speech (for example, Noun phrase, verb phrase, etc.), each level node after the second level node is obtained by the segmentation phrase of the upper level node according to the part of speech, until it is divided into the last level node of each node branch.
  • the first-level node is the comment itself
  • the second-level node is a participle phrase obtained by the comment according to the order of the corresponding participle and part of speech (for example, Noun phrase, verb phrase, etc.)
  • each level node after the second level node is obtained by the segmentation phrase of the upper level node according to the part of speech, until it is divided into the last level node of each node branch.
  • the participle phrase is the last level node of the node branch where it is located, and “I go to the playground to play football”, and the preset structure word segmentation tree is constructed as shown in Fig. 3. Show.
  • the node distance between each first keyword segmentation word (for example, a noun) and each second keyword segmentation word (for example, an adjective) is calculated based on the preset structure word segmentation tree, and the node distance is the first keyword segmentation word segmentation and the second
  • the number of nodes separated by the keyword participles respectively acquires a second keyword segmentation word having the smallest distance from each first keyword segmentation node, and the second keyword segmentation words with the smallest distance between each first keyword segmentation word and the node are in accordance with the comment.
  • the order in the composition constitutes the corresponding core viewpoint information.
  • FIG. 4 is a schematic flowchart of a method for recommending information in the application, and the method for recommending information includes the following steps:
  • Step S1 Before recommending the target object to the user, determining an approximate object corresponding to the target object according to a mapping relationship between the predetermined target object and the approximated object, and using the target object and the determined approximated object as the object to be analyzed;
  • the target object may be location information (such as restaurant location, library location or attraction location, etc.), item information (such as clothing accessories, books or daily necessities, etc.), network information (such as news, websites or audio and video, etc.), and financial management. Information (such as stocks, securities, insurance products, etc.).
  • the approximate object is an object similar to the target object. For example, if the target object is a restaurant location, the corresponding approximate object may be a food street location, a restaurant location, etc., and the target object is a library location, and the corresponding approximate object may be a variety of bookstores. Location, bookstore location, museum location, etc.
  • the target object and its corresponding approximation object are mapped in advance, and the mapping relationship between the target object and the approximation object is stored, for example, in the form of a mapping table.
  • the approximate object that has a mapping relationship with the target object is obtained according to the target object, and both the target object and the approximated object are used as subsequent objects to be analyzed.
  • step S2 the evaluation data of each object to be analyzed by each user in a preset time is obtained from a predetermined data source, and the evaluation data of each object to be analyzed is analyzed according to a predetermined analysis rule, and corresponding to each object to be analyzed is obtained.
  • Forward tag value the evaluation data of each object to be analyzed by each user in a preset time is obtained from a predetermined data source, and the evaluation data of each object to be analyzed is analyzed according to a predetermined analysis rule, and corresponding to each object to be analyzed is obtained.
  • the predetermined data source is, for example, a catering platform, a microblog website, a forum website, a life insurance server, etc.
  • the evaluation data includes a comment and a rating
  • each user is obtained from the predetermined data source to the waiting time within a preset time.
  • the evaluation data made by the analysis object is obtained, for example, from the predetermined data source, the evaluation data of the object to be analyzed by each user within six months is obtained.
  • the forward label value is a value in which the evaluation data is a variable, and the larger the evaluation amount in which the evaluation is high in the evaluation data, the larger the forward label value.
  • the predetermined analysis rule preferably includes:
  • evaluation data of the object to be analyzed is a score, and the score of the object to be analyzed is less than the preset score, determining that the object to be analyzed is an invalid object, or if the score is greater than or equal to the preset score, determining that the evaluation data is a positive information class And calculating a forward label value corresponding to the evaluation data of the object to be analyzed according to a preset calculation rule;
  • the evaluation data of the object to be analyzed is a comment, parsing the core viewpoint information corresponding to the comment, and using the pre-trained classifier to identify the information pointing category corresponding to the core viewpoint information (where the information pointing category includes the positive information category, Negative information class), if the information pointing category is a negative information class, determining that the object to be analyzed is an invalid object, or if the information pointing category is a positive information class, calculating the object to be analyzed according to a preset calculation rule The forward tag value corresponding to the evaluation data.
  • Step S3 if the forward label value is greater than or equal to a preset threshold, determining whether to recommend the target object to the user according to a predetermined recommendation algorithm;
  • Step S4 if yes, determining that the user is a related user of the target object, and recommending the target object to the user.
  • a preset threshold is preset.
  • the value of the forward label of the object to be analyzed is greater than or equal to the preset threshold, it indicates that the object to be analyzed is better evaluated or recognized by the user, and further considered according to a predetermined recommendation algorithm. Recommend the target object to the user, when the forward target of the object to be analyzed When the sign value is less than the preset threshold, the target object is not considered to be recommended to the user.
  • the predetermined recommendation algorithm may be: calculating a recommended value based on the forward label value, and determining, by the recommended value, whether to recommend the target object to the user, specifically, when the recommended value is greater than or equal to the preset recommended value,
  • the target object is recommended to the user, for example, the recommendation information of the target object is sent to the user, and the target object is not recommended to the user when the recommended value is less than the preset recommended value.
  • the present application also provides a computer readable storage medium having stored thereon a system for information recommendation, the step of implementing the above-described information recommendation method when the information recommendation system is executed by the processor.
  • the foregoing embodiment method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be through hardware, but in many cases, the former is better.
  • Implementation Based on such understanding, the technical solution of the present application, which is essential or contributes to the prior art, may be embodied in the form of a software product stored in a storage medium (such as ROM/RAM, disk,
  • the optical disc includes a number of instructions for causing a terminal device (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to perform the methods described in various embodiments of the present application.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Linguistics (AREA)
  • Artificial Intelligence (AREA)
  • Health & Medical Sciences (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • General Health & Medical Sciences (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

一种信息推荐的装置、方法、系统及计算机可读存储介质，信息推荐的装置包括存储器及处理器，存储器中存储有信息推荐的系统，信息推荐的系统被处理器执行时实现：在向用户推荐目标对象前，根据预定的目标对象与近似对象的映射关系确定目标对象对应的近似对象，并将目标对象和所确定的近似对象均作为待分析对象（S1）；从预定的数据源中获取各个用户在预设时间内对待分析对象所做的评价数据，根据预定的分析规则分析各个待分析对象的评价数据，得到各个待分析对象对应的正向标签值（S2）；若正向标签值大于等于预设阈值，则根据预定的推荐算法确定是否将该目标对象推荐给该用户（S3）；若是，则确定所述用户为所述目标对象的相关用户，则向用户推荐目标对象（S4）。上述信息推荐的系统能够准确地向用户推荐个性化信息。

Description

信息推荐的装置、方法、系统及计算机可读存储介质
优先权申明
本申请基于巴黎公约申明享有2017年08月20日递交的申请号为CN 201710715452.1、名称为“信息推荐的装置、方法及计算机可读存储介质”中国专利申请的优先权，该中国专利申请的整体内容以参考的方式结合在本申请中。
技术领域
本申请涉及通信技术领域，尤其涉及一种信息推荐的装置、方法、系统及计算机可读存储介质。
背景技术
目前的个性化信息推荐方案中，主要通过分析用户的历史行为得出用户的兴趣点，然后根据用户的兴趣点推送各种信息。通过用户所使用的社交媒体(例如微博、博客、论坛、播客等)可以获取到用户的历史行为，现有技术中通过分析用户的历史行为得到用户的正向偏好，例如用户A在微博上发布了较多的旅游及美食的文字、照片，则可以分析出用户A的正向偏好是喜欢旅游及美食。然而，一方面，用户的历史行为是有限的，基于用户上网历史行为向用户推荐个性化信息并不准确；另一方面，用户的兴趣是相对稳定的，而社会背景处于动态变化中(例如当前发生了热点新闻或时间等)，仅仅通过分析用户的历史行为的方式同样无法准确地向用户推荐个性化信息。
发明内容
本申请的目的在于提供一种信息推荐的装置、方法、系统及计算机可读存储介质，旨在准确地向用户推荐个性化信息。
为实现上述目的，本申请提供一种信息推荐的装置，所述信息推荐的装置包括存储器及与所述存储器连接的处理器，所述存储器中存储有可在所述处理器上运行的信息推荐的系统，所述信息推荐的系统被所述处理器执行时实现如下步骤：
S1，在向用户推荐目标对象之前，根据预定的目标对象与近似对象的映射关系确定所述目标对象对应的近似对象，并将所述目标对象和所确定的近似对象均作为待分析对象；
S2，从预定的数据源中获取各个用户在预设时间内对所述待分析对象所做的评价数据，根据预定的分析规则分析各个待分析对象的评价数据，得到各个待分析对象对应的正向标签值；
S3，若所述正向标签值大于等于预设阈值，则根据预定的推荐算法确定是否将该目标对象推荐给该用户；
S4，若是，则确定所述用户为所述目标对象的相关用户，向所述用户推 荐所述目标对象。
为实现上述目的，本申请还提供一种信息推荐的方法，所述信息推荐的方法包括：
S1，在向用户推荐目标对象之前，根据预定的目标对象与近似对象的映射关系确定所述目标对象对应的近似对象，并将所述目标对象和所确定的近似对象均作为待分析对象；
S2，从预定的数据源中获取各个用户在预设时间内对所述待分析对象所做的评价数据，根据预定的分析规则分析各个待分析对象的评价数据，得到各个待分析对象对应的正向标签值；
S3，若所述正向标签值大于等于预设阈值，则根据预定的推荐算法确定是否将该目标对象推荐给该用户；
S4，若是，则确定所述用户为所述目标对象的相关用户，向所述用户推荐所述目标对象。
为实现上述目的，本申请还提供一种信息推荐的系统，所述信息推荐的系统包括：
第一确定模块，用于在向用户推荐目标对象之前，根据预定的目标对象与近似对象的映射关系确定所述目标对象对应的近似对象，并将所述目标对象和所确定的近似对象均作为待分析对象；
分析模块，用于从预定的数据源中获取各个用户在预设时间内对所述待分析对象所做的评价数据，根据预定的分析规则分析各个待分析对象的评价数据，得到各个待分析对象对应的正向标签值；
第二确定模块，用于若所述正向标签值大于等于预设阈值，则根据预定的推荐算法确定是否将该目标对象推荐给该用户；
推荐模块，用于若是，则确定所述用户为所述目标对象的相关用户，向所述用户推荐所述目标对象。
本申请还提供一种计算机可读存储介质，所述计算机可读存储介质上存储有信息推荐的系统，所述信息推荐的系统被处理器执行时实现步骤：
S1，在向用户推荐目标对象之前，根据预定的目标对象与近似对象的映射关系确定所述目标对象对应的近似对象，并将所述目标对象和所确定的近似对象均作为待分析对象；
S2，从预定的数据源中获取各个用户在预设时间内对所述待分析对象所做的评价数据，根据预定的分析规则分析各个待分析对象的评价数据，得到各个待分析对象对应的正向标签值；
S3，若所述正向标签值大于等于预设阈值，则根据预定的推荐算法确定是否将该目标对象推荐给该用户；
S4，若是，则确定所述用户为所述目标对象的相关用户，向所述用户推荐所述目标对象。
本申请的有益效果是：本申请在向用户推荐目标对象之前，根据预定的映射关系确定目标对象对应的近似对象，并将目标对象和近似对象均作为待 分析对象；从预定的数据源中获取各个用户在预设时间内对待分析对象所做的评价数据，根据预定的分析规则分析各个待分析对象的评价数据，得到各个待分析对象对应的正向标签值；当正向标签值大于等于预设阈值，根据预定的推荐算法确定是否将该目标对象推荐给该用户，以将目标对象推荐给用户，本申请基于分析目标对象和近似对象的评价数据的方式向用户推荐目标对象，能够准确地向用户推荐个性化信息。
附图说明
图1为本申请各个实施例一可选的应用环境示意图；
图2是图1中信息推荐的装置一实施例的硬件架构的示意图；
图3为预设结构分词树的结构示意图；
图4为本申请信息推荐的方法一实施例的流程示意图。
具体实施方式
为了使本申请的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本申请进行进一步详细说明。应当理解，此处所描述的具体实施例仅用以解释本申请，并不用于限定本申请。基于本申请中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
需要说明的是，在本申请中涉及“第一”、“第二”等的描述仅用于描述目的，而不能理解为指示或暗示其相对重要性或者隐含指明所指示的技术特征的数量。由此，限定有“第一”、“第二”的特征可以明示或者隐含地包括至少一个该特征。另外，各个实施例之间的技术方案可以相互结合，但是必须是以本领域普通技术人员能够实现为基础，当技术方案的结合出现相互矛盾或无法实现时应当认为这种技术方案的结合不存在，也不在本申请要求的保护范围之内。
参阅图1，是本申请各个实施例一可选的应用环境示意图。在本实施例中，本申请可应用于包括，但不仅限于，信息推荐的装置1、用户终端2、网络3的应用环境中。信息推荐的装置1通过网络3与一个或多个用户终端2进行通信。
所述用户终端2包括，但不限于，任何一种可与用户通过键盘、鼠标、遥控器、触摸板或者声控设备等方式进行人机交互的电子产品，例如，个人计算机、平板电脑、智能手机、个人数字助理(Personal Digital Assistant，PDA)、游戏机、交互式网络电视(Internet Protocol Television,IPTV)、智能式穿戴式设备、导航装置等等的可移动设备，或者诸如数字TV、台式计算机、笔记本、服务器等等的固定终端。
其中，所述网络3可以是企业内部网(Intranet)、互联网(Internet)、全球移动通讯系统(Global System of Mobile communication，GSM)、宽带码分多址(Wideband Code Division Multiple Access，WCDMA)、4G网络、5G网 络、蓝牙(Bluetooth)、Wi-Fi等无线或有线网络。
所述信息推荐的装置1是一种能够按照事先设定或者存储的指令，自动进行数值计算和/或信息处理的设备。所述信息推荐的装置1可以是计算机、也可以是单个网络服务器、多个网络服务器组成的服务器组或者基于云计算的由大量主机或者网络服务器构成的云，其中云计算是分布式计算的一种，由一群松散耦合的计算机集组成的一个超级虚拟计算机。
参阅图2，是图1中信息推荐的装置1一可选的硬件架构的示意图，本实施例中，信息推荐的装置1可包括，但不仅限于，可通过系统总线相互通信连接的存储器11、处理器12、网络接口13。需要指出的是，图2仅示出了具有组件11-13的信息推荐的装置1，但是应理解的是，并不要求实施所有示出的组件，可以替代的实施更多或者更少的组件。
其中，所述存储器11至少包括一种类型的可读存储介质，所述可读存储介质包括闪存、硬盘、多媒体卡、卡型存储器(例如，SD或DX存储器等)、随机访问存储器(RAM)、静态随机访问存储器(SRAM)、只读存储器(ROM)、电可擦除可编程只读存储器(EEPROM)、可编程只读存储器(PROM)、磁性存储器、磁盘、光盘等。在一些实施例中，所述存储器11可以是所述信息推荐的装置1的内部存储单元，例如该信息推荐的装置1的硬盘或内存。在另一些实施例中，所述存储器11也可以是所述信息推荐的装置1的外部存储设备，例如该信息推荐的装置1上配备的插接式硬盘，智能存储卡(Smart Media Card,SMC)，安全数字(Secure Digital,SD)卡，闪存卡(Flash Card)等。当然，所述存储器11还可以既包括所述信息推荐的装置1的内部存储单元也包括其外部存储设备。本实施例中，所述存储器11通常用于存储安装于所述信息推荐的装置1的系统和各类应用软件，例如所述信息推荐的系统的程序代码等。此外，所述存储器11还可以用于暂时地存储已经输出或者将要输出的各类数据。
所述处理器12在一些实施例中可以是中央处理器(Central Processing Unit，CPU)、控制器、微控制器、微处理器、或其他数据处理芯片。该处理器12通常用于控制所述信息推荐的装置1的总体操作，例如执行与所述用户终端2进行通信相关的控制和处理等。本实施例中，所述处理器12用于运行所述存储器11中存储的程序代码或者处理数据，例如运行所述信息推荐的系统等。
所述网络接口13可包括无线网络接口或有线网络接口，该网络接口13通常用于在所述信息推荐的装置1与其他电子设备之间建立通信连接。本实施例中，所述网络接口13主要用于通过所述网络3将所述信息推荐的装置1与一个或多个所述用户终端2相连，以建立数据传输通道和通信连接。
所述信息推荐的系统存储在存储器11中，包括至少一个存储在存储器11中的计算机可读指令，该至少一个计算机可读指令可被处理器器12执行，以实现本申请各实施例的信息推荐的方法；以及，该至少一个计算机可读指令依据其各部分所实现的功能不同，可被划为不同的逻辑模块，包括第一确 定模块、分析模块、第二确定模块及推荐模块。
其中，上述信息推荐的系统被所述处理器12执行时实现如下步骤：
步骤S1，在向用户推荐目标对象之前，根据预定的目标对象与近似对象的映射关系确定所述目标对象对应的近似对象，并将所述目标对象和所确定的近似对象均作为待分析对象；
其中，目标对象可以是地点信息(例如餐厅地点、图书馆地点或景点地点等)、物品信息(例如衣服饰物、书籍或生活用品等)、网络信息(例如新闻、网站或音视频等)、理财信息(例如股票、证券、保险产品等)等。近似对象为与目标对象类似的对象，例如，目标对象为餐厅地点，则对应的近似对象可以是美食街地点、酒楼地点等，目标对象为图书馆地点，则对应的近似对象可以是各种书城的地点、书店地点、博物馆地点等。
本实施例中，预先将目标对象与其对应的近似对象进行映射，并存储目标对象与近似对象的映射关系，例如以映射表的形式进行存储。在向用户推荐目标对象之前，根据该目标对象获取与其存在映射关系的近似对象，将目标对象及近似对象均作为后续的待分析对象。
步骤S2，从预定的数据源中获取各个用户在预设时间内对所述待分析对象所做的评价数据，根据预定的分析规则分析各个待分析对象的评价数据，得到各个待分析对象对应的正向标签值；
本实施例中，预定的数据源例如为餐饮平台、微博网站、论坛网站、寿险服务器等，评价数据包括评论及评分，从预定的数据源中获取各个用户在预设时间内对所述待分析对象所做的评价数据，例如从预定的数据源中获取各个用户在六个月内对待分析对象所做的评价数据。正向标签值为以评价数据为变量的值，评价数据中评价高的评价量越大则正向标签值越大。
其中，预定的分析规则优选地包括：
若有待分析对象的评价数据为评分，且该待分析对象的评分小于预设评分，则确定该待分析对象为无效对象，或者，若大于等于预设评分，则确定该评价数据为正面信息类，并根据预设计算规则计算该待分析对象的评价数据对应的正向标签值；
若有待分析对象的评价数据为评论，则解析出该评论对应的核心观点信息，利用预先训练的分类器识别出所述核心观点信息对应的信息指向类别(其中，信息指向类别包括正面信息类、负面信息类)，若所述信息指向类别为负面信息类，则确定该待分析对象为无效对象，或者，若所述信息指向类别为正面信息类，则根据预设计算规则计算该待分析对象的评价数据对应的正向标签值。
步骤S3，若所述正向标签值大于等于预设阈值，则根据预定的推荐算法确定是否将该目标对象推荐给该用户；
步骤S4，若是，则确定所述用户为所述目标对象的相关用户，向所述用户推荐所述目标对象。
本实施例中，预定一预设阈值，当有待分析对象的正向标签值大于等于 预设阈值时，表明该待分析对象评价较好或者是被用户所认可，进一步根据预定的推荐算法考虑是否将该目标对象推荐给用户，当待分析对象的正向标签值均小于该预设阈值时，不考虑将该目标对象推荐给用户。
本实施例中，预定的推荐算法可以是基于正向标签值计算出推荐值，以该推荐值判定是否将目标对象推荐给用户，具体地，当该推荐值大于等于预设推荐值时将该目标对象推荐给用户，例如向用户发送该目标对象的推荐信息等，当该推荐值小于预设推荐值时不将该目标对象推荐给用户。
与现有技术相比，本实施例在向用户推荐目标对象之前，根据预定的映射关系确定目标对象对应的近似对象，并将目标对象和近似对象均作为待分析对象；从预定的数据源中获取各个用户在预设时间内对待分析对象所做的评价数据，根据预定的分析规则分析各个待分析对象的评价数据，得到各个待分析对象对应的正向标签值；当正向标签值大于等于预设阈值，根据预定的推荐算法确定是否将该目标对象推荐给该用户，以将目标对象推荐给用户，本实施例基于分析目标对象和近似对象的评价数据的方式向用户推荐目标对象，能够准确地向用户推荐个性化信息。
在一优选的实施例中，在上述实施例的基础上，所述信息推荐的系统被所述处理器12执行实现所述步骤S3时，具体包括：
若所述正向标签值大于等于预设阈值，则基于所述正向标签值及待分析对象的用户数据构建的推荐值计算公式计算得到所述目标对象的推荐值，在所述推荐值大于等于预设推荐值时，将所述目标对象推荐给该用户，其中，所述推荐值计算公式为：
P(o|u,t)＝λP(o|u)+(1-λ)P(o|δt)，
Figure PCTCN2017108761-appb-000001
其中，所述P(o|u,t)为时序t背景下用户u对目标对象或近似对象o的推荐值，所述λ为权重(λ为常量)，所述P(o|u)为用户u对目标对象或近似对象o的正向标签值，所述P(o|δt)为目标对象或近似对象o被用户u选择的概率，δt表示时序背景t下的话题分布，所述C＝{c1,c2,……,cn}为时序t背景下用户u的用户生成内容，n为用户生成内容的数量，所述sim(.)为一条用户生成内容(例如用户发布的微博或者论坛的评论等)与目标对象的相似度，所述o.w＝{w1,w2,……,wm}为目标对象o的关键字集合。以微博平台为例，可以简单理解为：
每个话题都一定“活跃”周期，假设这个周期为t，那么设用户u在这段时期内发布的微博集合为C＝{c1,c2,……,cn}，同时，设目标对象的关键字集合为o.w＝{w1,w2,……,wm}，则得到：
Figure PCTCN2017108761-appb-000002
其中，sim(.)表示一条微博与目标对象的相似度，优选地，本实施例采 用TF-IDF(term frequency–inverse document frequency)模型来计算所述相似度的值。
在一优选的实施例中，在上述实施例的基础上，上述的预设计算规则包括：计算出该用户针对该待分析对象做出的属于正面信息类的评价数据的第一数量，及该用户针对所有待分析对象做出的属于正面信息类的评价数据的第二数量；计算出所有用户针对该待分析对象做出的属于正面信息类的评价数据的第三数量，及所有用户针对所有待分析对象做出的属于正面信息类的评价数据的第四数量；将第一数量除以第二数量以得出第一正向标签参数，并将第三数量和第四数量代入预设公式
Figure PCTCN2017108761-appb-000003
以得出第二正向标签参数，所述
Figure PCTCN2017108761-appb-000004
为第二正向标签参数，所述B为第三数量，所述A为第四数量；将第一正向标签参数乘以第二正向标签参数，得出该待分析对象的评价数据对应的正向标签值。
优选地，所述分类器为支持向量机分类器，所述分类器的训练过程包括：
获取预设数量(例如10000个)正面信息类的核心观点信息样本(例如，平安健康险保障范围广、平安车险大品牌理赔快)，及预设数量(例如4000个)负面信息类的核心观点信息样本(例如，平安车险理赔慢服务差、平安理财产品没有承诺的高等)，将所有核心观点信息样本随机分成第一预设比例(例如，70％)的训练集和第二预设比例(例如，30％)的验证集，利用训练集训练分类器，并利用验证集验证训练的分类器的准确率，若准确率大于或者等于预设准确率(例如0.98)，则训练结束，或者，若准确率小于预设准确率，则增加正面信息类的核心观点信息样本数量及负面信息类的核心观点信息样本，并重新进行训练，直至训练的分类器的准确率大于或者等于预设准确率，训练结束。
在一优选的实施例中，在上述实施例的基础上，所述信息推荐的系统被所述处理器12执行实现所述解析出该评论对应的核心观点信息的步骤时，具体包括：
对该评论进行分词处理，并对该评论对应的各个分词进行词性标注；
根据该评论对应的各个分词的顺序及词性，将该评论对应的各个分词构建成预设结构分词树，基于该评论对应的各个分词构建成预设结构分词树解析出该评论对应的核心观点信息。
其中，分词包括字和词，例如对于评论“**推出的***产品很不错”，分词后的结果为“**”、“推出”、“的”、“***”、“很”、“不错”。
优选地，对评论进行分词处理包括：按预设类型标点符号(例如：“，”、“。”、“！”、“；”等)对该评论进行短句拆分，例如，从该评论起始位置至第一个预设类型标点符号之间的信息为一个短句；若评论结束位置无预设类型标点符号，则从倒数第一个预设类型标点符号至评论结束位置之间的信息为一个短句，且针对从第一个预设类型标点符号至倒数第一个预设类型标点 符号之间的信息，每两个预设类型标点符号之间的信息为一个短句；若评论结束位置有预设类型标点符号，则针对从第一个预设类型标点符号至倒数第一个预设类型标点符号之间的信息，每两个预设类型标点符号之间的信息为一个短句。
对拆分的每一个短句，采用长词优先原则继续进行分词：例如，长词优先原则指的是：对于一个需要分词的短语T1，先从第一个字A开始，从预存的词库找出一个由A起始的最长词语X1，然后从T1中剔除X1剩下T2，再对T2采用相同的切分原理，切分后的结果为“X1/X2/、、、、、、”，例如，评论“平安推出了尊宏人生产品”的切分结果为“平安”/“推出”/“了”/“尊宏人生”/“产品”。
优选地，对评论对应的各个分词进行词性标注包括：
根据通用字词典库中字和词分别与词性的映射关系(例如，通用字词典库中，操场对应的词性是名词)，及/或，预设的字和词分别与词性的映射关系(例如，预设的字和词分别与词性的映射关系中，操场对应的词性是常用名词)，确定各个短句的各个分词对应的词性，其中，预设的字和词分别与词性的映射关系的词性标注优先级高于通用字词典库中字和词分别与词性的映射关系。其中，词性包括：实词(名词、动词、形容词、数量词、代词等)、虚词(副词、介词、连词、助词、叹词、拟声词等)。
为各个分词标注对应的词性，例如，对于分词“了”、“来”、“着”、“过”、“的”、“地”、“得”、“似的”、“所”等等，标注其词性为助词；对于“非常安全”、“保本型”、“收益高”、“周期长”等，标注其词性为形容词。
优选地，根据该评论对应的各个分词的顺序及词性，将该评论对应的各个分词构建成预设结构分词树包括：
如图3所示，预设结构分词树包括多级节点，第一级节点为所述评论本身，第二级节点为由所述评论按照对应的分词的顺序及词性划分得到的分词短语(例如名词短语、动词短语等等)，第二级节点之后的每一级节点均是由上一级节点的分词短语按照词性继续划分得到的，直至划分至各节点分支的最后一级节点。在划分过程中，如果某一分词短语不能进一步划分，则该分词短语为所在的节点分支的最后一级节点，以“我去操场踢足球了”，构建的预设结构分词树如图3所示。
其中，基于构建的预设结构分词树，计算各个第一关键词性分词(例如，名词)与各个第二关键词性分词(例如，形容词)的节点距离，节点距离为第一关键词性分词与第二关键词性分词之间相隔的节点数，分别获取与各个第一关键词性分词节点距离最小的第二关键词性分词，将各个第一关键词性分词与节点距离最小的第二关键词性分词按照在该评论中的顺序组成对应的核心观点信息。
如图4所示，图4为本申请信息推荐的方法一实施例的流程示意图，该信息推荐的方法包括以下步骤：
步骤S1，在向用户推荐目标对象之前，根据预定的目标对象与近似对象的映射关系确定所述目标对象对应的近似对象，并将所述目标对象和所确定的近似对象均作为待分析对象；
其中，目标对象可以是地点信息(例如餐厅地点、图书馆地点或景点地点等)、物品信息(例如衣服饰物、书籍或生活用品等)、网络信息(例如新闻、网站或音视频等)、理财信息(例如股票、证券、保险产品等)等。近似对象为与目标对象类似的对象，例如，目标对象为餐厅地点，则对应的近似对象可以是美食街地点、酒楼地点等，目标对象为图书馆地点，则对应的近似对象可以是各种书城的地点、书店地点、博物馆地点等。
本实施例中，预先将目标对象与其对应的近似对象进行映射，并存储目标对象与近似对象的映射关系，例如以映射表的形式进行存储。在向用户推荐目标对象之前，根据该目标对象获取与其存在映射关系的近似对象，将目标对象及近似对象均作为后续的待分析对象。
步骤S2，从预定的数据源中获取各个用户在预设时间内对所述待分析对象所做的评价数据，根据预定的分析规则分析各个待分析对象的评价数据，得到各个待分析对象对应的正向标签值；
本实施例中，预定的数据源例如为餐饮平台、微博网站、论坛网站、寿险服务器等，评价数据包括评论及评分，从预定的数据源中获取各个用户在预设时间内对所述待分析对象所做的评价数据，例如从预定的数据源中获取各个用户在六个月内对待分析对象所做的评价数据。正向标签值为以评价数据为变量的值，评价数据中评价高的评价量越大则正向标签值越大。
其中，预定的分析规则优选地包括：
若有待分析对象的评价数据为评分，且该待分析对象的评分小于预设评分，则确定该待分析对象为无效对象，或者，若大于等于预设评分，则确定该评价数据为正面信息类，并根据预设计算规则计算该待分析对象的评价数据对应的正向标签值；
若有待分析对象的评价数据为评论，则解析出该评论对应的核心观点信息，利用预先训练的分类器识别出所述核心观点信息对应的信息指向类别(其中，信息指向类别包括正面信息类、负面信息类)，若所述信息指向类别为负面信息类，则确定该待分析对象为无效对象，或者，若所述信息指向类别为正面信息类，则根据预设计算规则计算该待分析对象的评价数据对应的正向标签值。
步骤S3，若所述正向标签值大于等于预设阈值，则根据预定的推荐算法确定是否将该目标对象推荐给该用户；
步骤S4，若是，则确定所述用户为所述目标对象的相关用户，向所述用户推荐所述目标对象。
本实施例中，预定一预设阈值，当有待分析对象的正向标签值大于等于预设阈值时，表明该待分析对象评价较好或者是被用户所认可，进一步根据预定的推荐算法考虑是否将该目标对象推荐给用户，当待分析对象的正向标 签值均小于该预设阈值时，不考虑将该目标对象推荐给用户。
本实施例中，预定的推荐算法可以是基于正向标签值计算出推荐值，以该推荐值判定是否将目标对象推荐给用户，具体地，当该推荐值大于等于预设推荐值时将该目标对象推荐给用户，例如向用户发送该目标对象的推荐信息等，当该推荐值小于预设推荐值时不将该目标对象推荐给用户。
本申请还提供一种计算机可读存储介质，所述计算机可读存储介质上存储有信息推荐的系统，所述信息推荐的系统被处理器执行时实现上述的信息推荐的方法的步骤。
上述本申请实施例序号仅仅为了描述，不代表实施例的优劣。
通过以上的实施方式的描述，本领域的技术人员可以清楚地了解到上述实施例方法可借助软件加必需的通用硬件平台的方式来实现，当然也可以通过硬件，但很多情况下前者是更佳的实施方式。基于这样的理解，本申请的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来，该计算机软件产品存储在一个存储介质(如ROM/RAM、磁碟、光盘)中，包括若干指令用以使得一台终端设备(可以是手机，计算机，服务器，空调器，或者网络设备等)执行本申请各个实施例所述的方法。
以上仅为本申请的优选实施例，并非因此限制本申请的专利范围，凡是利用本申请说明书及附图内容所作的等效结构或等效流程变换，或直接或间接运用在其他相关的技术领域，均同理包括在本申请的专利保护范围内。

Claims (20)

  1. 一种信息推荐的装置，其特征在于，所述信息推荐的装置包括存储器及与所述存储器连接的处理器，所述存储器中存储有可在所述处理器上运行的信息推荐的系统，所述信息推荐的系统被所述处理器执行时实现如下步骤：
    S1，在向用户推荐目标对象之前，根据预定的目标对象与近似对象的映射关系确定所述目标对象对应的近似对象，并将所述目标对象和所确定的近似对象均作为待分析对象；
    S2，从预定的数据源中获取各个用户在预设时间内对所述待分析对象所做的评价数据，根据预定的分析规则分析各个待分析对象的评价数据，得到各个待分析对象对应的正向标签值；
    S3，若所述正向标签值大于等于预设阈值，则根据预定的推荐算法确定是否将该目标对象推荐给该用户；
    S4，若是，则确定所述用户为所述目标对象的相关用户，向所述用户推荐所述目标对象。
  2. 根据权利要求1所述的信息推荐的装置，其特征在于，所述评价数据包括评分和评论，所述预定的分析规则包括：
    若有待分析对象的评价数据为评分，且该待分析对象的评分小于预设评分，则确定该待分析对象为无效对象，或者，若大于等于预设评分，则确定该评价数据为正面信息类，并根据预设计算规则计算该待分析对象的评价数据对应的正向标签值；
    若有待分析对象的评价数据为评论，则解析出该评论对应的核心观点信息，利用预先训练的分类器识别出所述核心观点信息对应的信息指向类别，若所述信息指向类别为负面信息类，则确定该待分析对象为无效对象，或者，若所述信息指向类别为正面信息类，则根据预设计算规则计算该待分析对象的评价数据对应的正向标签值。
  3. 根据权利要求2所述的信息推荐的装置，其特征在于，所述信息推荐的系统被所述处理器执行实现所述步骤S3时，具体包括：
    若所述正向标签值大于等于预设阈值，则基于所述正向标签值及待分析对象的用户数据构建的推荐值计算公式计算得到所述目标对象的推荐值，在所述推荐值大于等于预设推荐值时，将所述目标对象推荐给该用户，其中，所述推荐值计算公式为：
    P(o|u,t)＝λP(o|u)+(1-λ)P(o|δt)，
    Figure PCTCN2017108761-appb-100001
    其中，所述P(o|u,t)为时序t背景下用户u对目标对象或近似对象o的推荐值，所述λ为权重，所述P(o|u)为用户u对目标对象或近似对象o的正向标签值，所述P(o|δt)为目标对象或近似对象o被用户u选择的概率，所述δt表示时序背 景t下的话题分布，所述C＝{c1,c2,……,cn}为时序t背景下用户u的用户生成内容，所述n为用户生成内容的数量，所述sim(.)为一条用户生成内容与目标对象的相似度，所述o.w＝{w1,w2,……,wm}为目标对象o的关键字集合。
  4. 根据权利要求2所述的信息推荐的装置，其特征在于，所述预设计算规则包括：
    计算出该用户针对该待分析对象做出的属于正面信息类的评价数据的第一数量，及该用户针对所有待分析对象做出的属于正面信息类的评价数据的第二数量；
    计算出所有用户针对该待分析对象做出的属于正面信息类的评价数据的第三数量，及所有用户针对所有待分析对象做出的属于正面信息类的评价数据的第四数量；
    基于所述第一数量及第二数量获取第一正向标签参数，基于所述第三数量、第四数量及预设公式获取第二正向标签参数，并基于所述第一正向标签参数及第二正向标签参数得出该待分析对象的评价数据对应的正向标签值，其中，所述预设公式为：
    Figure PCTCN2017108761-appb-100002
    所述
    Figure PCTCN2017108761-appb-100003
    为第二正向标签参数，所述B为第三数量，所述A为第四数量。
  5. 根据权利要求2所述的信息推荐的装置，其特征在于，所述信息推荐的系统被所述处理器执行实现所述解析出该评论对应的核心观点信息的步骤时，具体包括：
    对该评论进行分词处理，并对该评论对应的各个分词进行词性标注；
    根据该评论对应的各个分词的顺序及词性，将该评论对应的各个分词构建成预设结构分词树，基于该评论对应的各个分词构建成预设结构分词树解析出该评论对应的核心观点信息。
  6. 根据权利要求5所述的信息推荐的装置，其特征在于，所述信息推荐的系统被所述处理器执行实现所述对该评论进行分词处理的步骤包括：
    按预设类型标点符号对该评论进行短句拆分，对拆分的每一个短句，采用长词优先原则继续进行分词；
    所述信息推荐的系统被所述处理器执行实现所述对该评论对应的各个分词进行词性标注的步骤包括：
    根据通用字词典库中字和词分别与词性的映射关系，及/或，预设的字和词分别与词性的映射关系，确定各个分词对应的词性，并进行标注，其中，预设的字和词分别与词性的映射关系的词性标注优先级高于通用字词典库中字和词分别与词性的映射关系。
  7. 根据权利要求6所述的信息推荐的装置，其特征在于，所述预设结构分词树包括多级节点，第一级节点为所述评论，第二级节点为由所述评论按照对应的分词的顺序及词性划分得到的分词短语，第二级节点之后的每一级节点均是由上一级节点的分词短语按照词性划分得到。
  8. 根据权利要求7所述的信息推荐的装置，其特征在于，所述信息推荐 的系统被所述处理器执行实现所述基于该评论对应的各个分词构建成预设结构分词树解析出该评论对应的核心观点信息的步骤包括：
    基于构建的预设结构分词树计算各个第一关键词性分词与各个第二关键词性分词的节点距离；分别获取与各个第一关键词性分词节点距离最小的第二关键词性分词，将各个第一关键词性分词与节点距离最小的第二关键词性分词按照在该评论中的顺序组成对应的核心观点信息。
  9. 一种信息推荐的方法，其特征在于，所述信息推荐的方法包括：
    S1，在向用户推荐目标对象之前，根据预定的目标对象与近似对象的映射关系确定所述目标对象对应的近似对象，并将所述目标对象和所确定的近似对象均作为待分析对象；
    S2，从预定的数据源中获取各个用户在预设时间内对所述待分析对象所做的评价数据，根据预定的分析规则分析各个待分析对象的评价数据，得到各个待分析对象对应的正向标签值；
    S3，若所述正向标签值大于等于预设阈值，则根据预定的推荐算法确定是否将该目标对象推荐给该用户；
    S4，若是，则确定所述用户为所述目标对象的相关用户，向所述用户推荐所述目标对象。
  10. 根据权利要求9所述的信息推荐的方法，其特征在于，所述评价数据包括评分和评论，所述预定的分析规则包括：
    若有待分析对象的评价数据为评分，且该待分析对象的评分小于预设评分，则确定该待分析对象为无效对象，或者，若大于等于预设评分，则确定该评价数据为正面信息类，并根据预设计算规则计算该待分析对象的评价数据对应的正向标签值；
    若有待分析对象的评价数据为评论，则解析出该评论对应的核心观点信息，利用预先训练的分类器识别出所述核心观点信息对应的信息指向类别，若所述信息指向类别为负面信息类，则确定该待分析对象为无效对象，或者，若所述信息指向类别为正面信息类，则根据预设计算规则计算该待分析对象的评价数据对应的正向标签值。
  11. 根据权利要求10所述的信息推荐的方法，其特征在于，所述步骤S3时，具体包括：
    若所述正向标签值大于等于预设阈值，则基于所述正向标签值及待分析对象的用户数据构建的推荐值计算公式计算得到所述目标对象的推荐值，在所述推荐值大于等于预设推荐值时，将所述目标对象推荐给该用户，其中，所述推荐值计算公式为：
    P(o|u,t)＝λP(o|u)+(1-λ)P(o|δt)，
    Figure PCTCN2017108761-appb-100004
    其中，所述P(o|u,t)为时序t背景下用户u对目标对象或近似对象o的推荐值，所述λ为权重，所述P(o|u)为用户u对目标对象或近似对象o的正向标签值，所述 P(o|δt)为目标对象或近似对象o被用户u选择的概率，所述δt表示时序背景t下的话题分布，所述C＝{c1,c2,……,cn}为时序t背景下用户u的用户生成内容，所述n为用户生成内容的数量，所述sim(.)为一条用户生成内容与目标对象的相似度，所述o.w＝{w1,w2,……,wm}为目标对象o的关键字集合。
  12. 根据权利要求10所述的信息推荐的方法，其特征在于，所述预设计算规则包括：
    计算出该用户针对该待分析对象做出的属于正面信息类的评价数据的第一数量，及该用户针对所有待分析对象做出的属于正面信息类的评价数据的第二数量；
    计算出所有用户针对该待分析对象做出的属于正面信息类的评价数据的第三数量，及所有用户针对所有待分析对象做出的属于正面信息类的评价数据的第四数量；
    基于所述第一数量及第二数量获取第一正向标签参数，基于所述第三数量、第四数量及预设公式获取第二正向标签参数，并基于所述第一正向标签参数及第二正向标签参数得出该待分析对象的评价数据对应的正向标签值，其中，所述预设公式为：
    Figure PCTCN2017108761-appb-100005
    所述
    Figure PCTCN2017108761-appb-100006
    为第二正向标签参数，所述B为第三数量，所述A为第四数量。
  13. 根据权利要求10所述的信息推荐的方法，其特征在于，所述解析出该评论对应的核心观点信息的步骤时，具体包括：
    对该评论进行分词处理，并对该评论对应的各个分词进行词性标注；
    根据该评论对应的各个分词的顺序及词性，将该评论对应的各个分词构建成预设结构分词树，基于该评论对应的各个分词构建成预设结构分词树解析出该评论对应的核心观点信息。
  14. 根据权利要求13所述的信息推荐的方法，其特征在于，所述对该评论进行分词处理的步骤包括：
    按预设类型标点符号对该评论进行短句拆分，对拆分的每一个短句，采用长词优先原则继续进行分词；
    所述信息推荐的系统被所述处理器执行实现所述对该评论对应的各个分词进行词性标注的步骤包括：
    根据通用字词典库中字和词分别与词性的映射关系，及/或，预设的字和词分别与词性的映射关系，确定各个分词对应的词性，并进行标注，其中，预设的字和词分别与词性的映射关系的词性标注优先级高于通用字词典库中字和词分别与词性的映射关系。
  15. 根据权利要求14所述的信息推荐的方法，其特征在于，所述预设结构分词树包括多级节点，第一级节点为所述评论，第二级节点为由所述评论按照对应的分词的顺序及词性划分得到的分词短语，第二级节点之后的每一级节点均是由上一级节点的分词短语按照词性划分得到。
  16. 根据权利要求15所述的信息推荐的方法，其特征在于，所述基于该评论对应的各个分词构建成预设结构分词树解析出该评论对应的核心观点信息的步骤包括：
    基于构建的预设结构分词树计算各个第一关键词性分词与各个第二关键词性分词的节点距离；分别获取与各个第一关键词性分词节点距离最小的第二关键词性分词，将各个第一关键词性分词与节点距离最小的第二关键词性分词按照在该评论中的顺序组成对应的核心观点信息。
  17. 一种信息推荐的系统，其特征在于，所述信息推荐的系统包括：
    第一确定模块，用于在向用户推荐目标对象之前，根据预定的目标对象与近似对象的映射关系确定所述目标对象对应的近似对象，并将所述目标对象和所确定的近似对象均作为待分析对象；
    分析模块，用于从预定的数据源中获取各个用户在预设时间内对所述待分析对象所做的评价数据，根据预定的分析规则分析各个待分析对象的评价数据，得到各个待分析对象对应的正向标签值；
    第二确定模块，用于若所述正向标签值大于等于预设阈值，则根据预定的推荐算法确定是否将该目标对象推荐给该用户；
    推荐模块，用于若是，则确定所述用户为所述目标对象的相关用户，向所述用户推荐所述目标对象。
  18. 根据权利要求17所述的信息推荐的系统，其特征在于，所述评价数据包括评分和评论，所述预定的分析规则包括：
    若有待分析对象的评价数据为评分，且该待分析对象的评分小于预设评分，则确定该待分析对象为无效对象，或者，若大于等于预设评分，则确定该评价数据为正面信息类，并根据预设计算规则计算该待分析对象的评价数据对应的正向标签值；
    若有待分析对象的评价数据为评论，则解析出该评论对应的核心观点信息，利用预先训练的分类器识别出所述核心观点信息对应的信息指向类别，若所述信息指向类别为负面信息类，则确定该待分析对象为无效对象，或者，若所述信息指向类别为正面信息类，则根据预设计算规则计算该待分析对象的评价数据对应的正向标签值。
  19. 根据权利要求18所述的信息推荐的系统，其特征在于，所述第二确定模块具体用于：
    若所述正向标签值大于等于预设阈值，则基于所述正向标签值及待分析对象的用户数据构建的推荐值计算公式计算得到所述目标对象的推荐值，在所述推荐值大于等于预设推荐值时，将所述目标对象推荐给该用户，其中，所述推荐值计算公式为：
    P(o|u,t)＝λP(o|u)+(1-λ)P(o|δt)，
    Figure PCTCN2017108761-appb-100007
    其中，所述P(o|u,t)为时序t背景下用户u对目标对象或近似对象o的推荐值，所述λ为 权重，所述P(o|u)为用户u对目标对象或近似对象o的正向标签值，所述P(o|δt)为目标对象或近似对象o被用户u选择的概率，所述δt表示时序背景t下的话题分布，所述C＝{c1,c2,……,cn}为时序t背景下用户u的用户生成内容，所述n为用户生成内容的数量，所述sim(.)为一条用户生成内容与目标对象的相似度，所述o.w＝{w1,w2,……,wm}为目标对象o的关键字集合。
  20. 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质上存储有信息推荐的系统，所述信息推荐的系统被处理器执行时实现如权利要求9-16任一项所述的信息推荐的方法的步骤。
PCT/CN2017/108761 2017-08-20 2017-10-31 信息推荐的装置、方法、系统及计算机可读存储介质 Ceased WO2019037258A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201710715452.1A CN107526800B (zh) 2017-08-20 2017-08-20 信息推荐的装置、方法及计算机可读存储介质
CN201710715452.1 2017-08-20

Publications (1)

Publication Number Publication Date
WO2019037258A1 true WO2019037258A1 (zh) 2019-02-28

Family

ID=60681606

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/108761 Ceased WO2019037258A1 (zh) 2017-08-20 2017-10-31 信息推荐的装置、方法、系统及计算机可读存储介质

Country Status (2)

Country Link
CN (1) CN107526800B (zh)
WO (1) WO2019037258A1 (zh)

Families Citing this family (24)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110110267B (zh) * 2018-01-25 2024-07-16 北京京东尚科信息技术有限公司 提取对象特征、搜索对象的方法和装置
CN108182608B (zh) * 2018-01-30 2021-04-02 重庆金融资产交易所有限责任公司 电子装置、产品推荐方法和计算机可读存储介质
CN108961069A (zh) * 2018-05-24 2018-12-07 中国平安人寿保险股份有限公司 电子装置、个人投保单数据处理方法和计算机存储介质
CN109033211A (zh) * 2018-07-01 2018-12-18 东莞市华睿电子科技有限公司 一种基于城市的街头艺术表演点的互动方法
CN109002507A (zh) * 2018-07-01 2018-12-14 东莞市华睿电子科技有限公司 一种大数据平台与品牌车型相结合的社交用户推荐方法
CN109033232B (zh) * 2018-07-01 2021-12-28 东莞市华睿电子科技有限公司 一种云平台与共享设备相结合的社交用户推荐方法
CN109002506A (zh) * 2018-07-01 2018-12-14 东莞市华睿电子科技有限公司 一种由品牌车型关注引起的信息推荐方法
CN109325796B (zh) * 2018-08-13 2023-09-26 中国平安人寿保险股份有限公司 潜在用户筛选方法、装置、计算机设备及存储介质
CN109242545A (zh) * 2018-08-20 2019-01-18 中国平安人寿保险股份有限公司 转发链接的评分方法、装置、计算机设备和存储介质
CN110866768B (zh) * 2018-08-27 2024-04-16 阿里巴巴集团控股有限公司 一种评价方法及装置
CN109670853B (zh) * 2018-09-25 2023-09-19 平安科技（深圳）有限公司 用户特征数据的确定方法、装置、设备及可读存储介质
CN109241444B (zh) * 2018-10-11 2024-10-11 平安科技（深圳）有限公司 基于状态机的内容推荐方法、装置、设备和存储介质
CN109639561B (zh) * 2018-10-25 2022-03-04 北京达佳互联信息技术有限公司 基于信息反馈的分享方法、装置、电子设备及存储介质
CN109857922A (zh) * 2019-01-18 2019-06-07 深圳壹账通智能科技有限公司 数据评测模型建模方法、装置、计算机设备及存储介质
CN110263248B (zh) * 2019-05-21 2023-11-28 平安科技（深圳）有限公司 一种信息推送方法、装置、存储介质和服务器
CN110610072B (zh) * 2019-09-06 2021-10-01 武汉达梦数据库有限公司 一种数据鉴权方法及装置
CN110866180B (zh) * 2019-10-12 2022-07-29 平安国际智慧城市科技股份有限公司 资源推荐方法、服务器及存储介质
TWI742450B (zh) * 2019-10-16 2021-10-11 財團法人工業技術研究院 企劃書的自動化產生系統及其方法
CN114595671B (zh) * 2020-12-07 2024-08-20 腾讯科技（深圳）有限公司 推荐信息生成方法及装置、存储介质、计算设备
CN113434772B (zh) * 2021-07-09 2023-07-25 支付宝(杭州)信息技术有限公司 推荐处理方法及装置
CN114385926B (zh) * 2021-12-02 2024-11-15 山东师范大学 攻击环境下降低无用信息比重的鲁棒性推荐方法及系统
CN114549034A (zh) * 2021-12-24 2022-05-27 珠海大横琴科技发展有限公司 一种费用定价的方法和装置
CN114547396B (zh) * 2022-02-28 2025-06-24 亿咖通(湖北)技术有限公司 待推送信息召回处理方法、装置、设备、介质及程序产品
CN116362761B (zh) * 2023-03-06 2024-04-05 北京三维天地科技股份有限公司 基于数据聚合推荐算法的检验检测机构推荐方法及系统

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103136257A (zh) * 2011-11-30 2013-06-05 阿里巴巴集团控股有限公司 信息提供方法及其装置
CN103970857A (zh) * 2014-05-07 2014-08-06 百度在线网络技术（北京）有限公司 推荐内容确定系统和方法
CN103970858A (zh) * 2014-05-07 2014-08-06 百度在线网络技术（北京）有限公司 推荐内容确定系统和方法
CN104699711A (zh) * 2013-12-09 2015-06-10 华为技术有限公司 一种推荐方法及服务器
US20150169727A1 (en) * 2012-09-03 2015-06-18 Sony Corporation Information processing apparatus, information processing method, and system
US20170186069A1 (en) * 2015-12-29 2017-06-29 International Business Machines Corporation Credibility enhancement for online comments and recommendations

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6317722B1 (en) * 1998-09-18 2001-11-13 Amazon.Com, Inc. Use of electronic shopping carts to generate personal recommendations
US9009589B2 (en) * 2007-03-30 2015-04-14 Google Inc. Conversion of portable program modules for constrained displays
CN101329683A (zh) * 2008-07-25 2008-12-24 华为技术有限公司 推荐系统及方法
CN103853789A (zh) * 2012-12-06 2014-06-11 深圳市点行天下科技有限公司 一种用于向用户推荐信息的方法和设备
CN104090912B (zh) * 2014-06-10 2017-03-15 腾讯科技（深圳）有限公司 信息推送方法及装置
CN106021586B (zh) * 2016-06-06 2020-02-18 腾讯科技（北京）有限公司 一种信息处理方法及服务器

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103136257A (zh) * 2011-11-30 2013-06-05 阿里巴巴集团控股有限公司 信息提供方法及其装置
US20150169727A1 (en) * 2012-09-03 2015-06-18 Sony Corporation Information processing apparatus, information processing method, and system
CN104699711A (zh) * 2013-12-09 2015-06-10 华为技术有限公司 一种推荐方法及服务器
CN103970857A (zh) * 2014-05-07 2014-08-06 百度在线网络技术（北京）有限公司 推荐内容确定系统和方法
CN103970858A (zh) * 2014-05-07 2014-08-06 百度在线网络技术（北京）有限公司 推荐内容确定系统和方法
US20170186069A1 (en) * 2015-12-29 2017-06-29 International Business Machines Corporation Credibility enhancement for online comments and recommendations

Also Published As

Publication number Publication date
CN107526800B (zh) 2019-03-26
CN107526800A (zh) 2017-12-29

Similar Documents

Publication Publication Date Title
WO2019037258A1 (zh) 信息推荐的装置、方法、系统及计算机可读存储介质
US10630798B2 (en) Artificial intelligence based method and apparatus for pushing news
US10642938B2 (en) Artificial intelligence based method and apparatus for constructing comment graph
US10891322B2 (en) Automatic conversation creator for news
US10747771B2 (en) Method and apparatus for determining hot event
US20230177360A1 (en) Surfacing unique facts for entities
CN104750798B (zh) 一种应用程序的推荐方法和装置
CN107077470A (zh) 聚焦的语义分类
CN107368515A (zh) 应用程序页面推荐方法及系统
CN110717038B (zh) 对象分类方法及装置
US20230076387A1 (en) Systems and methods for providing a comment-centered news reader
Wilson et al. Linguistic steganography on twitter: hierarchical language modeling with manual interaction
US9251141B1 (en) Entity identification model training
US10699078B2 (en) Comment-centered news reader
CN114925185A (zh) 交互方法、模型的训练方法、装置、设备及介质
CN111369148A (zh) 对象指标的监测方法、电子装置及存储介质
WO2019041528A1 (zh) 新闻情感方向判断方法、电子设备及计算机可读存储介质
CN111382354A (zh) 一种信息推荐方法、装置、设备及存储介质
CN109981712B (zh) 用于推送信息的方法和装置
CN112148958A (zh) 用于信息推荐的方法、设备和计算机存储介质
CN115099368B (zh) 一种用于计算篇章级文档相似度的方法及可读存储介质
US9946765B2 (en) Building a domain knowledge and term identity using crowd sourcing
JP2017091436A (ja) 特徴語選択装置
CN116431912A (zh) 用户画像推送方法及装置
JP2020154790A (ja) 情報処理装置、情報処理方法、及びプログラム

Legal Events

Date Code Title Description
NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS (EPO FORM 1205A DATED 30.09.2020)

122 Ep: pct application non-entry in european phase

Ref document number: 17922216

Country of ref document: EP

Kind code of ref document: A1