WO2019041520A1 - 基于社交数据的金融产品推荐方法、电子装置及介质 - Google Patents

基于社交数据的金融产品推荐方法、电子装置及介质 Download PDF

Info

Publication number
WO2019041520A1
WO2019041520A1 PCT/CN2017/108796 CN2017108796W WO2019041520A1 WO 2019041520 A1 WO2019041520 A1 WO 2019041520A1 CN 2017108796 W CN2017108796 W CN 2017108796W WO 2019041520 A1 WO2019041520 A1 WO 2019041520A1
Authority
WO
WIPO (PCT)
Prior art keywords
financial
user
demand
latest
vocabulary
Prior art date
Application number
PCT/CN2017/108796
Other languages
English (en)
French (fr)
Inventor
毕野
肖京
王建明
Original Assignee
平安科技（深圳）有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 平安科技（深圳）有限公司 filed Critical 平安科技（深圳）有限公司
Publication of WO2019041520A1 publication Critical patent/WO2019041520A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—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
    • G06F40/279—Recognition of textual entities
    • G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/30—Semantic analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/01—Social networking

Definitions

  • the present application relates to the field of computer technology, and in particular, to a financial product recommendation method, an electronic device, and a medium based on social data.
  • the recommendation system of the existing financial field is usually based on user static image data to implement financial product recommendation of a specific user.
  • user static image data cannot update the data in time, there is a poor timeliness, which leads to the poor timeliness of the recommendation system in the existing financial field, and the recommendation effect is poor.
  • the purpose of the present application is to provide a financial product recommendation method, an electronic device and a medium based on social data, aiming at improving the timeliness of financial product recommendation.
  • a first aspect of the present application provides an electronic device including a memory, a processor, and a social data-based financial product recommendation system stored on the memory and operable on the processor
  • the social data based financial product recommendation system is implemented by the processor to implement the following steps:
  • A1 Obtain social data of the latest preset time period of the user from a preset data source
  • D1 sorting all vocabulary vectors according to the similarity of the keywords according to the similarity from high to low, and selecting a preset number of vocabulary vectors that are ranked first;
  • E1 acquiring a corresponding keyword of the preset number of vocabulary vectors, and determining a user's latest financial demand theme according to the financial demand theme marked by the corresponding keyword;
  • the second aspect of the present application provides a financial product recommendation method based on social data
  • the financial product recommendation methods based on social data include:
  • A2 Obtain social data of the latest preset time period of the user from a preset data source
  • C2 Perform similarity calculation on each vocabulary vector and a preset keyword respectively, and determine that the keyword with the highest similarity with each vocabulary vector is a corresponding keyword of the vocabulary vector, and the keyword is pre-marked with different finances. Demand theme;
  • E2 acquiring a corresponding keyword of the preset number of vocabulary vectors, and determining a user's latest financial demand theme according to the financial demand topic marked by the corresponding keyword;
  • a third aspect of the present application provides a computer readable storage medium storing a social data based financial product recommendation system, the social data based financial product recommendation system being executable by at least one processor, Taking the at least one processor to perform the following steps:
  • A3. Obtain social data of the latest preset time period of the user from a preset data source
  • E3 Obtain a corresponding keyword of the preset number of vocabulary vectors, and determine a user's latest financial demand theme according to the financial demand topic marked by the corresponding keyword;
  • the social data-based financial product recommendation method, the electronic device and the medium proposed by the present application determine the latest financial demand theme of the user by analyzing the social data of the latest preset time period of the user, and based on the latest financial demand theme of the user The user makes recommendations for the corresponding financial products.
  • financial product recommendation can unearth the user's own demand preferences for financial products expressed through social data.
  • the analysis of the user's financial needs is more precise, which can greatly improve the timeliness and effectiveness of financial product recommendations.
  • FIG. 1 is a schematic diagram of an operating environment of a preferred embodiment of a financial product recommendation system 10 of the present application;
  • FIG. 2 is a schematic flowchart diagram of an embodiment of a method for recommending a financial product based on social data.
  • 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 operating environment of a preferred embodiment of the financial product recommendation system 10 of the present application.
  • the financial product recommendation system 10 is installed and operated in the electronic device 1.
  • the electronic device 1 may include, but is not limited to, a memory 11, a processor 12, and a display 13.
  • Figure 1 shows only the electronic device 1 with 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 comprises at least one type of readable storage medium, which in some embodiments may be an internal storage unit of the electronic device 1, such as a hard disk or memory of the electronic device 1.
  • the memory 11 may also be an external storage device of the electronic device 1 in other embodiments, such as a plug-in hard disk equipped on the electronic device 1, a smart memory card (SMC), and a secure digital device. (Secure Digital, SD) card, flash card, etc.
  • the memory 11 may also include both an internal storage unit of the electronic device 1 and an external storage device.
  • the memory 11 is configured to store application software installed on the electronic device 1 and various types of data, such as program codes of the financial product recommendation system 10 and the like.
  • the memory 11 can also be used to temporarily store data that has been output or is about to be output.
  • the processor 12 may be a central processor (Central) in some embodiments A processing unit (CPU), a microprocessor or other data processing chip for running program code or processing data stored in the memory 11, for example, executing the financial product recommendation system 10 or the like.
  • Central central processor
  • CPU central processing unit
  • microprocessor microprocessor or other data processing chip for running program code or processing data stored in the memory 11, for example, executing the financial product recommendation system 10 or the like.
  • the display 13 in some embodiments may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch sensor, or the like.
  • the display 13 is used to display information processed in the electronic device 1 and a user interface for displaying visualization, such as the determined user's latest financial demand theme, recommended financial products, and the like.
  • the components 11-13 of the electronic device 1 communicate with one another via a system bus.
  • the social data based financial product recommendation system 10 includes at least one computer readable instructions stored in the memory 11, the at least one computer readable instructions being executable by the processor 12 to implement various embodiments of the present application.
  • the above-mentioned social data-based financial product recommendation system 10 is executed by the processor 12 to implement the following steps:
  • Step S1 Acquire social data of the latest preset time period of the user from a preset data source.
  • the social data of the latest preset time period of the user is obtained from the preset data source, for example, from the user's QQ, Weibo, WeChat, snowball, Social software such as Oriental Fortune obtains the user's latest preset time period, such as the last 3 months, 6 months of social data, including but not limited to the user's initiative to send articles, circle of friends, etc., and the user posted to others Comments on content, forwarding content, and more.
  • the social data of the user's latest preset time period is acquired, the amount of acquired data that is preset according to the social type may be different, for example, due to the present embodiment.
  • the financial data of the financial class is highly important, and the social data of the financial APP of the last 3 months can be obtained by default after triggering the social data acquisition, while the other types default to the social data of the last 2 months.
  • step S2 the social data is segmented, and each participle calculation is converted into a vocabulary vector.
  • the social data may be segmented by a preset word segmentation method.
  • the word segmentation method of the string matching may be used to perform word segmentation processing on the social data, such as a forward maximum matching method, and the character string in one information is segmented from left to right, that is, the social data is left to right.
  • Several consecutive characters in the table match the vocabulary, if the match, a word is segmented; or, the reverse maximum matching method, the character string in a message is segmented from right to left, that is, from the social data
  • the end of the matching scan starts, and several consecutive characters in the information text of the word to be distinguished are matched with the vocabulary from right to left.
  • Word segmentation is a word segmentation method for machine speech judgment. It uses syntactic information and semantic information to deal with ambiguity phenomena to segment words.
  • the statistical word segmentation method can also be used for word segmentation processing. From the current user's historical search record or the public user's historical search record, according to the statistics of the phrase, some two adjacent words appear to have more frequent frequencies. These two adjacent words can be used as a phrase to perform word segmentation.
  • word segmentation can also be performed based on the Chinese word segmentation tool NLPIR, which is not limited herein.
  • the word vector model word2vec is used to convert each word segmentation into a vocabulary vector, wherein word2vec is a tool for converting words into a vector form, which can simplify the processing of the text content into a vector space.
  • the vector operation calculates the similarity in the vector space to represent the semantic similarity of the text.
  • word2vec can be used to simplify the processing of text content into vector operations in K-dimensional vector space, and the similarity in vector space can be used to represent the semantic similarity of text. Therefore, the word vector output by word2vec can be used to do things like clustering, finding synonyms, part of speech analysis, etc. Moreover, word2vec is very efficient.
  • Step S3 performing similarity calculation on each vocabulary vector and the preset keyword respectively, and determining that the keyword with the highest similarity with each vocabulary vector is the corresponding keyword of the vocabulary vector, and the keywords are pre-marked differently. Financial needs theme.
  • a theme dictionary for characterizing different financial needs of customers is first constructed, and keywords in the dictionary are marked according to different financial demand topics.
  • the different financial needs of the mark include, but are not limited to, investment, wealth management, insurance, health, loan financial topics and so on.
  • keywords such as “borrowing money”, “borrowing”, and “hands-on” in the dictionary can be marked as “loan” financial topics.
  • the converted lexical vector may be similarly calculated with the keywords in the dictionary marked with different financial demand topics, for example, by word2vec
  • the keywords in the dictionary marked with different financial needs topics are converted into vector forms.
  • the similarity calculation of several lexical vectors and keywords can be simplified into vector operations in vector space, and the similarity in vector space can be calculated.
  • a plurality of vocabulary vectors may be separately calculated for each keyword in a keyword having a different financial demand theme from a preset mark.
  • each vocabulary vector of the converted vocabulary vectors can be similarly calculated with the keywords to determine the keywords with the highest similarity with each vocabulary vector. Is the corresponding keyword of the vocabulary vector.
  • vocabulary vector A can be used for each of “borrowing money”, “borrowing”, “loan”, “free money”, “revenue” and “financial management”.
  • the key words are respectively calculated for similarity.
  • the similarity between the vocabulary vector A and each keyword can be calculated, and the keyword with the highest similarity with the vocabulary vector A is selected as the corresponding keyword of the vocabulary vector A, such as the vocabulary vector A and "borrowing money", "borrowing", "
  • the similarities between loans, free money, income, and financial management are 90%, 80%, 40%, 30%, 30%, and 30%, respectively, and the keyword with the highest similarity to vocabulary vector A is “ If you borrow money, you will use "borrowing money” as the corresponding keyword of vocabulary vector A.
  • step S4 all vocabulary vectors are sorted according to the similarity from high to low according to the similar keyword similarity, and a preset number of vocabulary vectors with the highest ranking are selected.
  • the similarity between the vocabulary vector a and the corresponding keyword “borrowing” is 99%, the similarity between the vocabulary vector b and the corresponding keyword “free money” is 98%, and the similarity between the vocabulary vector c and the corresponding keyword “borrowing” For 97%, the similarity between the vocabulary vector d and the corresponding keyword “return” is 96%, then the order of similarity from high to low is a, b, c, d.
  • a preset number for example, 50
  • Step S5 Acquire a corresponding keyword of the preset number of vocabulary vectors, and determine a user's latest financial demand theme according to the financial demand topic marked by the corresponding keyword.
  • the financial demand theme can be marked according to keywords.
  • the vocabulary vector with the highest similarity to the keywords in the dictionary such as TOP50
  • the corresponding keywords according to the selected TOP50 vocabulary vector may be selected.
  • the subject of the financial needs tagged to determine the user's latest financial needs topic For example, the number or proportion of different financial demand topics in the financial demand topic marked by the corresponding keyword of the selected TOP50 vocabulary vector may be counted, and the number or the top financial demand theme with the highest number or the highest proportion may be selected.
  • selecting the number or the highest proportion of the TOP5 financial demand theme as the user's latest financial demand theme that is, representing the real-time hotspot demand of the user's latest time period.
  • Step S6 recommending the corresponding financial product to the user based on the latest financial demand theme of the user.
  • the topic model algorithm and the collaborative filtering algorithm implement the user's financial product recommendation, such as ranking recommendations for each financial product.
  • the recommendation priority of the financial product that meets the latest financial demand theme of the user is increased, so that the user is preferentially recommended to meet the latest financial needs of the user. Themed financial products.
  • the latest financial demand theme of the TOP5 of the user's recent time period is the real-time hotspot demand, and the financial product order that meets the TOP5 real-time hotspot demand is improved, thereby realizing the latest demand according to the user.
  • the preference for dynamically adjusting financial product recommendations has improved the timeliness of financial product recommendations.
  • the embodiment determines the latest financial demand theme of the user by analyzing the social data of the latest preset time period of the user, and recommends the corresponding financial product to the user based on the latest financial demand theme of the user. .
  • financial product recommendation can unearth the user's own demand preferences for financial products expressed through social data. The analysis of the user's financial needs is more precise, which can greatly improve the timeliness and effectiveness of financial product recommendations.
  • the step of determining the latest financial requirement theme of the user according to the financial requirement theme marked by the corresponding keyword in the step S3 includes:
  • the scores of different financial demand topics are calculated according to the preset scoring rules, and the financial demand with the highest score is selected.
  • the theme is the subject of the user's latest financial needs.
  • the vocabulary vector with the highest degree of similarity to the keyword in the dictionary, such as the TOP50 is selected, and the user is determined according to the financial demand topic marked by the corresponding keyword of the selected TOP50 vocabulary vector.
  • the number or proportion of different financial demand topics in the financial demand topic marked by the corresponding keywords of the selected TOP50 vocabulary vector can be counted.
  • the scores of different financial demand topics are calculated according to a preset scoring rule, and the selection is further selected.
  • the highest-rated financial needs theme is the subject of the user's latest financial needs.
  • the first weight value corresponding to the number or proportion of each different financial demand topic may be preset, and the second weight value corresponding to the similarity of each vocabulary vector and the corresponding keyword is based on the first weight value,
  • the comprehensive calculation of the two weights gives the scores of different financial demand topics.
  • the first weight value and the second weight value are rootable According to the actual application, the proportion of the theme of financial demand is focused on or the similarity between the vocabulary vector and the corresponding keyword is set separately, and the user's latest financial demand theme can be determined more accurately.
  • FIG. 2 is a schematic flowchart of a method for recommending a financial product based on social data according to an embodiment of the present invention.
  • the method for recommending a financial product based on social data includes the following steps:
  • Step S10 Acquire social data of the latest preset time period of the user from a preset data source.
  • the social data of the latest preset time period of the user is obtained from the preset data source, for example, from the user's QQ, Weibo, WeChat, snowball, Social software such as Oriental Fortune obtains the user's latest preset time period, such as the last 3 months, 6 months of social data, including but not limited to the user's initiative to send articles, circle of friends, etc., and the user posted to others Comments on content, forwarding content, and more.
  • the social data of the user's latest preset time period is acquired, the amount of acquired data that is preset according to the social type may be different, for example, due to the present embodiment.
  • the financial data of the financial class is highly important, and the social data of the financial APP of the last 3 months can be obtained by default after triggering the social data acquisition, while the other types default to the social data of the last 2 months.
  • Step S20 segmenting the social data, and converting each word segmentation into a vocabulary vector.
  • the social data may be segmented by a preset word segmentation method.
  • the word segmentation method of the string matching may be used to perform word segmentation processing on the social data, such as a forward maximum matching method, and the character string in one information is segmented from left to right, that is, the social data is left to right.
  • Several consecutive characters in the table match the vocabulary, if the match, a word is segmented; or, the reverse maximum matching method, the character string in a message is segmented from right to left, that is, from the social data
  • the end of the matching scan starts, and several consecutive characters in the information text of the word to be distinguished are matched with the vocabulary from right to left.
  • Word lexical segmentation can also be used to classify each piece of information. Word lexical segmentation is a segmentation method for machine speech judgment. It uses syntactic information and semantic information to deal with ambiguity phenomena to segment words. The statistical word segmentation method can also be used for word segmentation processing. From the current user's historical search record or the public user's historical search record, according to the statistics of the phrase, some two adjacent words appear to have more frequent frequencies. These two adjacent words can be used as a phrase to perform word segmentation. In addition, word segmentation can also be performed based on the Chinese word segmentation tool NLPIR, which is not limited herein.
  • word2vec is used to divide each score.
  • the word calculation is converted into a lexical vector, where word2vec is a tool for converting words into vector form, which can simplify the processing of text content into vector operations in vector space, calculate the similarity in vector space, and represent text semantics. Similarity on the top.
  • word2vec can be used to simplify the processing of text content into vector operations in K-dimensional vector space, and the similarity in vector space can be used to represent the semantic similarity of text. Therefore, the word vector output by word2vec can be used to do things like clustering, finding synonyms, part of speech analysis, etc. Moreover, word2vec is very efficient.
  • Step S30 Perform similarity calculation on each vocabulary vector and the preset keyword respectively, and determine that the keyword with the highest similarity with each vocabulary vector is the corresponding keyword of the vocabulary vector, and the keywords are pre-marked differently. Financial needs theme.
  • a theme dictionary for characterizing different financial needs of customers is first constructed, and keywords in the dictionary are marked according to different financial demand topics.
  • the different financial needs of the mark include, but are not limited to, investment, wealth management, insurance, health, loan financial topics and so on.
  • keywords such as “borrowing money”, “borrowing”, and “hands-on” in the dictionary can be marked as “loan” financial topics.
  • the converted lexical vector may be similarly calculated with the keywords in the dictionary marked with different financial demand topics, for example, by word2vec
  • the keywords in the dictionary marked with different financial needs topics are converted into vector forms.
  • the similarity calculation of several lexical vectors and keywords can be simplified into vector operations in vector space, and the similarity in vector space can be calculated.
  • a plurality of vocabulary vectors may be separately calculated for each keyword in a keyword having a different financial demand theme from a preset mark.
  • each vocabulary vector of the converted vocabulary vectors can be similarly calculated with the keywords to determine the keywords with the highest similarity with each vocabulary vector. Is the corresponding keyword of the vocabulary vector.
  • the vocabulary vector A can be similarly calculated for each of the keywords “borrowing money”, “borrowing”, “loan”, “free money”, “revenue”, and “financial wealth”.
  • the similarity between the vocabulary vector A and each keyword can be calculated, and the keyword with the highest similarity with the vocabulary vector A is selected as the corresponding keyword of the vocabulary vector A, such as the vocabulary vector A and "borrowing money", “borrowing", "
  • the similarities between loans, free money, income, and financial management are 90%, 80%, 40%, 30%, 30%, and 30%, respectively, and the keyword with the highest similarity to vocabulary vector A is “ If you borrow money, you will use "borrowing money” as the corresponding keyword of vocabulary vector A.
  • Step S40 according to the similar keyword similarity, all the vocabulary vectors are according to the similarity Sort from high to low, picking out the preset number of vocabulary vectors that are ranked first.
  • the similarity between the vocabulary vector a and the corresponding keyword “borrowing” is 99%, the similarity between the vocabulary vector b and the corresponding keyword “free money” is 98%, and the similarity between the vocabulary vector c and the corresponding keyword “borrowing” For 97%, the similarity between the vocabulary vector d and the corresponding keyword “return” is 96%, then the order of similarity from high to low is a, b, c, d.
  • a preset number for example, 50
  • Step S50 Acquire a corresponding keyword of the preset number of vocabulary vectors, and determine a user's latest financial demand theme according to the financial demand topic marked by the corresponding keyword.
  • the financial demand theme can be marked according to keywords.
  • the vocabulary vector with the highest similarity to the keywords in the dictionary such as TOP50
  • the corresponding keywords according to the selected TOP50 vocabulary vector may be selected.
  • the subject of the financial needs tagged to determine the user's latest financial needs topic For example, the number or proportion of different financial demand topics in the financial demand topic marked by the corresponding keyword of the selected TOP50 vocabulary vector may be counted, and the number or the top financial demand theme with the highest number or the highest proportion may be selected.
  • selecting the number or the highest proportion of the TOP5 financial demand theme as the user's latest financial demand theme that is, representing the real-time hotspot demand of the user's latest time period.
  • Step S60 recommending the corresponding financial product to the user based on the latest financial demand theme of the user.
  • the user's financial product recommendation is realized through matrix analysis algorithm, LDA topic model algorithm and collaborative filtering algorithm, such as Sort and recommend each financial product.
  • the recommendation priority of the financial product that meets the latest financial demand theme of the user is increased, so that the user is preferentially recommended to meet the latest financial needs of the user.
  • Themed financial products For example, in the financial products that are preset to the user, the latest financial demand theme of the TOP5 of the user's recent time period is the real-time hotspot demand, and the full-time hotspot demand is raised.
  • the TOP5 real-time hotspot demand for financial product order thus realizing the dynamic adjustment of financial product recommendations according to the user's latest demand preferences, and improving the timeliness of financial product recommendation.
  • the embodiment determines the latest financial demand theme of the user by analyzing the social data of the latest preset time period of the user, and recommends the corresponding financial product to the user based on the latest financial demand theme of the user. .
  • financial product recommendation can unearth the user's own demand preferences for financial products expressed through social data. The analysis of the user's financial needs is more precise, which can greatly improve the timeliness and effectiveness of financial product recommendations.
  • the step of determining, according to the financial requirement theme marked by the corresponding keyword in the step S30, the user's latest financial requirement theme includes:
  • the scores of different financial demand topics are calculated according to the preset scoring rules, and the financial demand with the highest score is selected.
  • the theme is the subject of the user's latest financial needs.
  • the vocabulary vector with the highest degree of similarity to the keyword in the dictionary, such as the TOP50 is selected, and the user is determined according to the financial demand topic marked by the corresponding keyword of the selected TOP50 vocabulary vector.
  • the number or proportion of different financial demand topics in the financial demand topic marked by the corresponding keywords of the selected TOP50 vocabulary vector can be counted.
  • the scores of different financial demand topics are calculated according to a preset scoring rule, and the selection is further selected.
  • the highest-rated financial needs theme is the subject of the user's latest financial needs.
  • the first weight value corresponding to the number or proportion of each different financial demand topic may be preset, and the second weight value corresponding to the similarity of each vocabulary vector and the corresponding keyword is based on the first weight value,
  • the comprehensive calculation of the two weights gives the scores of different financial demand topics.
  • the first weight value and the second weight value may be respectively set according to the proportion of the theme of the financial demand in the actual application or the similarity between the vocabulary vector and the corresponding keyword, and the user may be more accurately determined. The latest financial needs theme.
  • the present application also provides a computer readable storage medium storing a social product based financial product recommendation system, the social number based
  • the financial product recommendation system may be executed by at least one processor to cause the at least one processor to perform the steps of the social data-based financial product recommendation method as in the above embodiment, the social data-based financial product recommendation method
  • the specific implementation processes of steps S10, S20, and S30 are as described above, and are not described herein again.
  • the foregoing embodiment method can be implemented by means of software plus a necessary general hardware platform, and can also be implemented by hardware, but in many cases, the former is A better implementation.
  • 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)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Marketing (AREA)
  • Artificial Intelligence (AREA)
  • Computational Linguistics (AREA)
  • Strategic Management (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • General Business, Economics & Management (AREA)
  • Economics (AREA)
  • Data Mining & Analysis (AREA)
  • Development Economics (AREA)
  • Finance (AREA)
  • Technology Law (AREA)
  • Accounting & Taxation (AREA)
  • Computing Systems (AREA)
  • Human Resources & Organizations (AREA)
  • Primary Health Care (AREA)
  • Tourism & Hospitality (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)

Abstract

一种基于社交数据的金融产品推荐方法、电子装置及介质，该方法包括：从预设的数据源获取用户最新预设时间段的社交数据（S10）；对所述社交数据进行分词，并将各个分词计算转换为词汇向量（S20）；将每一词汇向量与预设的关键词分别进行相似度计算，确定与每一词汇向量相似度最高的关键词为所述词汇向量的对应关键词（S30）；根据对应关键词相似度将所有词汇向量按所述相似度由高到低进行排序，挑选出排序靠前的预设数量的词汇向量（S40）；获取所述预设数量的词汇向量的对应关键词，并根据所述对应关键词所标记的金融需求主题确定用户的最新金融需求主题（S50）；基于该用户的最新金融需求主题对该用户进行相应金融产品的推荐（S60）。所述方法提高金融产品推荐的时效性及效果。

Description

基于社交数据的金融产品推荐方法、电子装置及介质
本申请基于巴黎公约申明享有2017年8月31日递交的申请号为CN 201710773595.8、名称为“基于社交数据的金融产品推荐方法、电子装置及介质”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
本申请涉及计算机技术领域，尤其涉及一种基于社交数据的金融产品推荐方法、电子装置及介质。
背景技术
现有金融领域的推荐系统通常是以用户静态画像数据作为基础来实现特定用户的金融产品推荐。然而，由于用户静态画像数据无法及时更新数据，存在较差的时效性，从而导致现有金融领域的推荐系统的时效性较差，推荐效果较差。
发明内容
本申请的目的在于提供一种基于社交数据的金融产品推荐方法、电子装置及介质，旨在提高金融产品推荐的时效性。
为实现上述目的，本申请第一方面提供一种电子装置，所述电子装置包括存储器、处理器及存储在所述存储器上并可在所述处理器上运行的基于社交数据的金融产品推荐系统，所述基于社交数据的金融产品推荐系统被所述处理器执行时实现如下步骤：
A1、从预设的数据源获取用户最新预设时间段的社交数据；
B1、对所述社交数据进行分词，并将各个分词计算转换为词汇向量；
C1、将每一词汇向量与预设的关键词分别进行相似度计算，确定与每一词汇向量相似度最高的关键词为所述词汇向量的对应关键词，所述关键词预先标记有不同金融需求主题；
D1、根据对应关键词相似度将所有词汇向量按所述相似度由高到低进行排序，挑选出排序靠前的预设数量的词汇向量；
E1、获取所述预设数量的词汇向量的对应关键词，并根据所述对应关键词所标记的金融需求主题确定用户的最新金融需求主题；
F1、基于该用户的最新金融需求主题对该用户进行相应金融产品的推荐。
本申请第二方面提供一种基于社交数据的金融产品推荐方法，所 述基于社交数据的金融产品推荐方法包括：
A2、从预设的数据源获取用户最新预设时间段的社交数据；
B2、对所述社交数据进行分词，并将各个分词计算转换为词汇向量；
C2、将每一词汇向量与预设的关键词分别进行相似度计算，确定与每一词汇向量相似度最高的关键词为所述词汇向量的对应关键词，所述关键词预先标记有不同金融需求主题；
D2、根据对应关键词相似度将所有词汇向量按所述相似度由高到低进行排序，挑选出排序靠前的预设数量的词汇向量；
E2、获取所述预设数量的词汇向量的对应关键词，并根据所述对应关键词所标记的金融需求主题确定用户的最新金融需求主题；
F2、基于该用户的最新金融需求主题对该用户进行相应金融产品的推荐。
本申请第三方面提供一种计算机可读存储介质，所述计算机可读存储介质存储有基于社交数据的金融产品推荐系统，所述基于社交数据的金融产品推荐系统可被至少一个处理器执行，以使所述至少一个处理器执行如下步骤：
A3、从预设的数据源获取用户最新预设时间段的社交数据；
B3、对所述社交数据进行分词，并将各个分词计算转换为词汇向量；
C3、将每一词汇向量与预设的关键词分别进行相似度计算，确定与每一词汇向量相似度最高的关键词为所述词汇向量的对应关键词，所述关键词预先标记有不同金融需求主题；
D3、根据对应关键词相似度将所有词汇向量按所述相似度由高到低进行排序，挑选出排序靠前的预设数量的词汇向量；
E3、获取所述预设数量的词汇向量的对应关键词，并根据所述对应关键词所标记的金融需求主题确定用户的最新金融需求主题；
F3、基于该用户的最新金融需求主题对该用户进行相应金融产品的推荐。
本申请提出的基于社交数据的金融产品推荐方法、电子装置及介质，通过分析挖掘用户最新预设时间段的社交数据，确定该用户的最新金融需求主题，并基于该用户的最新金融需求主题对该用户进行相应金融产品的推荐。由于引入实时动态的社交数据，来预测用户对金融产品的需求偏好，并根据用户对金融产品的需求偏好进行金融产品推荐，能挖掘出用户本身通过社交数据所表达的对金融产品的需求偏好，对于用户金融需求的分析更加精准，能极大地提高金融产品推荐的时效性及效果。
附图说明
图1为本申请金融产品推荐系统10较佳实施例的运行环境示意图；
图2为本申请基于社交数据的金融产品推荐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。
本申请提供一种基于社交数据的金融产品推荐系统。请参阅图1，是本申请金融产品推荐系统10较佳实施例的运行环境示意图。
在本实施例中，所述的金融产品推荐系统10安装并运行于电子装置1中。该电子装置1可包括，但不仅限于，存储器11、处理器12及显示器13。图1仅示出了具有组件11-13的电子装置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。
所述存储器11至少包括一种类型的可读存储介质，所述存储器11在一些实施例中可以是所述电子装置1的内部存储单元，例如该电子装置1的硬盘或内存。所述存储器11在另一些实施例中也可以是所述电子装置1的外部存储设备，例如所述电子装置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。进一步地，所述存储器11还可以既包括所述电子装置1的内部存储单元也包括外部存储设备。所述存储器11用于存储安装于所述电子装置1的应用软件及各类数据，例如所述金融产品推荐系统10的程序代码等。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数据。
所述处理器12在一些实施例中可以是一中央处理器(Central  Processing Unit,CPU)，微处理器或其他数据处理芯片，用于运行所述存储器11中存储的程序代码或处理数据，例如执行所述金融产品推荐系统10等。
所述显示器13在一些实施例中可以是LED显示器、液晶显示器、触控式液晶显示器以及OLED(Organic Light-Emitting Diode，有机发光二极管)触摸器等。所述显示器13用于显示在所述电子装置1中处理的信息以及用于显示可视化的用户界面，例如确定出的用户最新金融需求主题、推荐的金融产品等。所述电子装置1的部件11-13通过系统总线相互通信。
基于社交数据的金融产品推荐系统10包括至少一个存储在所述存储器11中的计算机可读指令，该至少一个计算机可读指令可被所述处理器12执行，以实现本申请各实施例。
其中，上述基于社交数据的金融产品推荐系统10被所述处理器12执行时实现如下步骤：
步骤S1，从预设的数据源获取用户最新预设时间段的社交数据。
本实施例中，在接收到对用户进行金融产品的推荐请求时，首先从预设的数据源获取用户最新预设时间段的社交数据，例如从用户的QQ、微博、微信、雪球、东方财富等社交软件上获取用户最新预设时间段如最近3个月、6个月的社交数据，该社交数据包括但不限于用户主动发出的文章、朋友圈等内容，以及用户对其他人发布内容的评论、转发内容，等等。
进一步地，由于是对用户进行金融产品的推荐，因此，在获取用户最新预设时间段的社交数据时，可根据社交类型不同预先设定的获取的数据量不同，例如，由于在本实施例中金融类社交数据的重要程度高，则可在触发社交数据获取后默认获取最近3个月的金融类APP的社交数据，而其他类型则默认为获取最近2个月的社交数据等。
步骤S2，对所述社交数据进行分词，并将各个分词计算转换为词汇向量。
在获取到用户最新预设时间段的社交数据后，可利用预设的分词方式对所述社交数据进行分词。例如，可利用字符串匹配的分词方法对所述社交数据进行分词处理，如正向最大匹配法，把一个信息中的字符串从左至右来分词，即从左到右将所述社交数据中的几个连续字符与词表匹配，如果匹配上，则切分出一个词；或者，反向最大匹配法，把一个信息中的字符串从右至左来分词，即从所述社交数据的末端开始匹配扫描，从右至左将待分词的信息文本中的几个连续字符与词表匹配，如果匹配上，则切分出一个词；或者，最短路径分词法，一个信息中的字符串里面要求切出的词数是最少的；或者，双向最大匹配法，正反向同时进行分词匹配。还可利用词义分词法对各个信息 进行分词处理，词义分词法是一种机器语音判断的分词方法，利用句法信息和语义信息来处理歧义现象来分词。还可利用统计分词法对各个信息进行分词处理，从当前用户的历史搜索记录或大众用户的历史搜索记录中，根据词组的统计，会统计有些两个相邻的字出现的频率较多，则可将这两个相邻的字作为词组来进行分词。此外，还可基于中文分词工具NLPIR进行分词，在此不做限定。
对所述社交数据进行分词后，采用词向量模型word2vec将各个分词计算转换为词汇向量，其中，word2vec是一个将单词转换成向量形式的工具，可以把对文本内容的处理简化为向量空间中的向量运算，计算出向量空间上的相似度，来表示文本语义上的相似度。例如，word2vec通过训练，可以把对文本内容的处理简化为K维向量空间中的向量运算，而向量空间上的相似度可以用来表示文本语义上的相似度。因此，word2vec输出的词向量可以被用来做如聚类、找同义词、词性分析等等工作，而且，word2vec非常高效。
步骤S3，将每一词汇向量与预设的关键词分别进行相似度计算，确定与每一词汇向量相似度最高的关键词为所述词汇向量的对应关键词，所述关键词预先标记有不同金融需求主题。
本实施例中，首先构建表征客户不同金融需求主题词典，并依据不同金融需求主题对词典中的关键词进行标记。其中，标记的不同金融需求主题包括但不限于投资、理财、保险、健康、贷款金融主题等等。例如，对词典中的“借钱”、“借款”、“手头拮据”等关键词，可标记为“贷款”金融主题。
在对用户的社交数据进行分词，并将各个分词计算转换为词汇向量后，可将转换的若干词汇向量与词典中标记有不同金融需求主题的关键词进行相似度计算，例如，可通过word2vec将词典中标记有不同金融需求主题的关键词转换成向量形式，这样，可将若干词汇向量与关键词的相似度计算简化为向量空间中的向量运算，计算出向量空间上的相似度，即可用来表示若干词汇向量与关键词在文本语义上的相似度。具体地，可将若干词汇向量与预设的标记有不同金融需求主题的关键词中每一关键词分别进行相似度计算。
例如，若金融需求主题词典中的金融需求主题包括“贷款”和“理财”金融主题，标记“贷款”主题的关键词包括“借钱”、“借款”及“贷款”，标记“理财”主题的关键词包括“闲钱”、“收益”及“理财”。对用户的社交数据进行分词、转换成若干词汇向量后，可将转换得到的若干词汇向量中每一词汇向量与关键词分别进行相似度计算，以确定与每一词汇向量相似度最高的关键词为所述词汇向量的对应关键词。例如，针对词汇向量A，可将词汇向量A与“借钱”、“借款”、“贷款”、“闲钱”、“收益”及“理财”中的每一个关 键词分别进行相似度计算。可计算出词汇向量A与每个关键词的相似度，选择与词汇向量A相似度最高的关键词作为词汇向量A的对应关键词，如词汇向量A与“借钱”、“借款”、“贷款”、“闲钱”、“收益”、“理财”的相似度分别为90％、80％、40％、30％、30％、30％，则与词汇向量A相似度最高的关键词为“借钱”，则将“借钱”作为词汇向量A的对应关键词。
步骤S4，根据对应关键词相似度将所有词汇向量按所述相似度由高到低进行排序，挑选出排序靠前的预设数量的词汇向量。
例如，若词汇向量a与对应关键词“借款”的相似度为99％，词汇向量b与对应关键词“闲钱”的相似度为98％，词汇向量c与对应关键词“借款”的相似度为97％，词汇向量d与对应关键词“收益”的相似度为96％，则按相似度从高到低排序依次为a、b、c、d。以此，可根据相似度计算结果挑选出排序靠前即与关键词的相似度最高的预设数量(如50个)词汇向量。
步骤S5，获取所述预设数量的词汇向量的对应关键词，并根据所述对应关键词所标记的金融需求主题确定用户的最新金融需求主题。
获取所述预设数量(如50个)的词汇向量的对应关键词，并可根据预设数量(如50个)词汇向量的对应关键词所标记的金融需求主题确定该用户的最新金融需求主题。例如，挑选出的50个词汇向量的对应关键词中有10个“借钱”、10个“贷款”、20个“闲钱”和10个“收益”，则可根据关键词标记的金融需求主题确定50个词汇向量中包含“贷款”金融需求主题的词汇有20个，包含“理财”金融需求主题的词汇有30个，“理财”金融需求主题的词汇数量比“贷款”金融需求主题的词汇数量多，则可以此确定该用户的最新金融需求主题为“理财”金融需求主题。
计算出若干词汇向量与词典中关键词的相似度之后，可挑选出与词典中关键词的相似度最高的预设数量如TOP50的词汇向量，并根据挑选出的TOP50的词汇向量的对应关键词所标记的金融需求主题来确定该用户的最新金融需求主题。例如，可统计挑选出的TOP50的词汇向量的对应关键词所标记的金融需求主题中各个不同金融需求主题的数量或所占比例，选择数量或所占比例最高的一个或前几个金融需求主题作为该用户的最新金融需求主题，例如，选择数量或所占比例最高的TOP5的金融需求主题作为该用户的最新金融需求主题，即代表了该用户最新时间段的实时热点需求。
步骤S6，基于该用户的最新金融需求主题对该用户进行相应金融产品的推荐。
在预设的推荐给用户的若干金融产品中，通常是以用户静态画像数据和购买金融产品行为数据作为基础，通过矩阵分析算法、LDA 主题模型算法以及协同过滤算法实现用户的金融产品推荐，如对各个金融产品进行排序推荐。本实施例中，在预设推荐给该用户若干金融产品的基础上，将符合该用户的最新金融需求主题的金融产品的推荐优先级提高，以向该用户优先推荐符合该用户的最新金融需求主题的金融产品。例如，在预设排序推荐给用户的各个金融产品中，融合用户最近时间段的TOP5的最新金融需求主题即实时热点需求，提升满足TOP5实时热点需求的金融产品顺序，从而实现根据用户的最新需求偏好动态调整金融产品的推荐，提升了金融产品推荐的时效性。
与现有技术相比，本实施例通过分析挖掘用户最新预设时间段的社交数据，确定该用户的最新金融需求主题，并基于该用户的最新金融需求主题对该用户进行相应金融产品的推荐。由于引入实时动态的社交数据，来预测用户对金融产品的需求偏好，并根据用户对金融产品的需求偏好进行金融产品推荐，能挖掘出用户本身通过社交数据所表达的对金融产品的需求偏好，对于用户金融需求的分析更加精准，能极大地提高金融产品推荐的时效性及效果。
在一可选的实施例中，在上述实施例的基础上，所述步骤S3中根据所述对应关键词所标记的金融需求主题确定用户的最新金融需求主题的步骤包括：
确定预设数量的词汇向量的对应关键词所标记的金融需求主题中各个不同金融需求主题所占的数量；
根据预设数量的词汇向量中各个词汇向量与对应关键词的相似度，以及各个不同金融需求主题所占的数量按预设评分规则计算出各个不同金融需求主题的评分，选择评分最高的金融需求主题作为该用户的最新金融需求主题。
本实施例中，在挑选出与词典中关键词的相似度最高的预设数量如TOP50的词汇向量，并根据挑选出的TOP50的词汇向量的对应关键词所标记的金融需求主题来确定该用户的最新金融需求主题时，可统计挑选出的TOP50的词汇向量的对应关键词所标记的金融需求主题中各个不同金融需求主题的数量或所占比例。进一步地，还可根据TOP50的词汇向量中各个词汇向量与对应关键词的相似度，以及各个不同金融需求主题所占的数量或比例按预设评分规则计算出各个不同金融需求主题的评分，选择评分最高的金融需求主题作为该用户的最新金融需求主题。例如，可预先设定各个不同金融需求主题的数量或所占比例所对应的第一权重值，各个词汇向量与对应关键词的相似度所对应的第二权重值，基于第一权重值、第二权重值综合计算得到各个不同金融需求主题的评分。其中，第一权重值、第二权重值可根 据实际应用中是侧重于金融需求主题的占比或侧重于词汇向量与对应关键词的相似度来分别进行设定，能更加准确地确定出用户的最新金融需求主题。
如图2所示，图2为本申请基于社交数据的金融产品推荐方法一实施例的流程示意图，该基于社交数据的金融产品推荐方法包括以下步骤：
步骤S10，从预设的数据源获取用户最新预设时间段的社交数据。
本实施例中，在接收到对用户进行金融产品的推荐请求时，首先从预设的数据源获取用户最新预设时间段的社交数据，例如从用户的QQ、微博、微信、雪球、东方财富等社交软件上获取用户最新预设时间段如最近3个月、6个月的社交数据，该社交数据包括但不限于用户主动发出的文章、朋友圈等内容，以及用户对其他人发布内容的评论、转发内容，等等。
进一步地，由于是对用户进行金融产品的推荐，因此，在获取用户最新预设时间段的社交数据时，可根据社交类型不同预先设定的获取的数据量不同，例如，由于在本实施例中金融类社交数据的重要程度高，则可在触发社交数据获取后默认获取最近3个月的金融类APP的社交数据，而其他类型则默认为获取最近2个月的社交数据等。
步骤S20，对所述社交数据进行分词，并将各个分词计算转换为词汇向量。
在获取到用户最新预设时间段的社交数据后，可利用预设的分词方式对所述社交数据进行分词。例如，可利用字符串匹配的分词方法对所述社交数据进行分词处理，如正向最大匹配法，把一个信息中的字符串从左至右来分词，即从左到右将所述社交数据中的几个连续字符与词表匹配，如果匹配上，则切分出一个词；或者，反向最大匹配法，把一个信息中的字符串从右至左来分词，即从所述社交数据的末端开始匹配扫描，从右至左将待分词的信息文本中的几个连续字符与词表匹配，如果匹配上，则切分出一个词；或者，最短路径分词法，一个信息中的字符串里面要求切出的词数是最少的；或者，双向最大匹配法，正反向同时进行分词匹配。还可利用词义分词法对各个信息进行分词处理，词义分词法是一种机器语音判断的分词方法，利用句法信息和语义信息来处理歧义现象来分词。还可利用统计分词法对各个信息进行分词处理，从当前用户的历史搜索记录或大众用户的历史搜索记录中，根据词组的统计，会统计有些两个相邻的字出现的频率较多，则可将这两个相邻的字作为词组来进行分词。此外，还可基于中文分词工具NLPIR进行分词，在此不做限定。
对所述社交数据进行分词后，采用词向量模型word2vec将各个分 词计算转换为词汇向量，其中，word2vec是一个将单词转换成向量形式的工具，可以把对文本内容的处理简化为向量空间中的向量运算，计算出向量空间上的相似度，来表示文本语义上的相似度。例如，word2vec通过训练，可以把对文本内容的处理简化为K维向量空间中的向量运算，而向量空间上的相似度可以用来表示文本语义上的相似度。因此，word2vec输出的词向量可以被用来做如聚类、找同义词、词性分析等等工作，而且，word2vec非常高效。
步骤S30，将每一词汇向量与预设的关键词分别进行相似度计算，确定与每一词汇向量相似度最高的关键词为所述词汇向量的对应关键词，所述关键词预先标记有不同金融需求主题。
本实施例中，首先构建表征客户不同金融需求主题词典，并依据不同金融需求主题对词典中的关键词进行标记。其中，标记的不同金融需求主题包括但不限于投资、理财、保险、健康、贷款金融主题等等。例如，对词典中的“借钱”、“借款”、“手头拮据”等关键词，可标记为“贷款”金融主题。
在对用户的社交数据进行分词，并将各个分词计算转换为词汇向量后，可将转换的若干词汇向量与词典中标记有不同金融需求主题的关键词进行相似度计算，例如，可通过word2vec将词典中标记有不同金融需求主题的关键词转换成向量形式，这样，可将若干词汇向量与关键词的相似度计算简化为向量空间中的向量运算，计算出向量空间上的相似度，即可用来表示若干词汇向量与关键词在文本语义上的相似度。具体地，可将若干词汇向量与预设的标记有不同金融需求主题的关键词中每一关键词分别进行相似度计算。
例如，若金融需求主题词典中的金融需求主题包括“贷款”和“理财”金融主题，标记“贷款”主题的关键词包括“借钱”、“借款”及“贷款”，标记“理财”主题的关键词包括“闲钱”、“收益”及“理财”。对用户的社交数据进行分词、转换成若干词汇向量后，可将转换得到的若干词汇向量中每一词汇向量与关键词分别进行相似度计算，以确定与每一词汇向量相似度最高的关键词为所述词汇向量的对应关键词。例如，针对词汇向量A，可将词汇向量A与“借钱”、“借款”、“贷款”、“闲钱”、“收益”及“理财”中的每一个关键词分别进行相似度计算。可计算出词汇向量A与每个关键词的相似度，选择与词汇向量A相似度最高的关键词作为词汇向量A的对应关键词，如词汇向量A与“借钱”、“借款”、“贷款”、“闲钱”、“收益”、“理财”的相似度分别为90％、80％、40％、30％、30％、30％，则与词汇向量A相似度最高的关键词为“借钱”，则将“借钱”作为词汇向量A的对应关键词。
步骤S40，根据对应关键词相似度将所有词汇向量按所述相似度 由高到低进行排序，挑选出排序靠前的预设数量的词汇向量。
例如，若词汇向量a与对应关键词“借款”的相似度为99％，词汇向量b与对应关键词“闲钱”的相似度为98％，词汇向量c与对应关键词“借款”的相似度为97％，词汇向量d与对应关键词“收益”的相似度为96％，则按相似度从高到低排序依次为a、b、c、d。以此，可根据相似度计算结果挑选出排序靠前即与关键词的相似度最高的预设数量(如50个)词汇向量。
步骤S50，获取所述预设数量的词汇向量的对应关键词，并根据所述对应关键词所标记的金融需求主题确定用户的最新金融需求主题。
获取所述预设数量(如50个)的词汇向量的对应关键词，并可根据预设数量(如50个)词汇向量的对应关键词所标记的金融需求主题确定该用户的最新金融需求主题。例如，挑选出的50个词汇向量的对应关键词中有10个“借钱”、10个“贷款”、20个“闲钱”和10个“收益”，则可根据关键词标记的金融需求主题确定50个词汇向量中包含“贷款”金融需求主题的词汇有20个，包含“理财”金融需求主题的词汇有30个，“理财”金融需求主题的词汇数量比“贷款”金融需求主题的词汇数量多，则可以此确定该用户的最新金融需求主题为“理财”金融需求主题。
计算出若干词汇向量与词典中关键词的相似度之后，可挑选出与词典中关键词的相似度最高的预设数量如TOP50的词汇向量，并根据挑选出的TOP50的词汇向量的对应关键词所标记的金融需求主题来确定该用户的最新金融需求主题。例如，可统计挑选出的TOP50的词汇向量的对应关键词所标记的金融需求主题中各个不同金融需求主题的数量或所占比例，选择数量或所占比例最高的一个或前几个金融需求主题作为该用户的最新金融需求主题，例如，选择数量或所占比例最高的TOP5的金融需求主题作为该用户的最新金融需求主题，即代表了该用户最新时间段的实时热点需求。
步骤S60，基于该用户的最新金融需求主题对该用户进行相应金融产品的推荐。
在预设的推荐给用户的若干金融产品中，通常是以用户静态画像数据和购买金融产品行为数据作为基础，通过矩阵分析算法、LDA主题模型算法以及协同过滤算法实现用户的金融产品推荐，如对各个金融产品进行排序推荐。本实施例中，在预设推荐给该用户若干金融产品的基础上，将符合该用户的最新金融需求主题的金融产品的推荐优先级提高，以向该用户优先推荐符合该用户的最新金融需求主题的金融产品。例如，在预设排序推荐给用户的各个金融产品中，融合用户最近时间段的TOP5的最新金融需求主题即实时热点需求，提升满 足TOP5实时热点需求的金融产品顺序，从而实现根据用户的最新需求偏好动态调整金融产品的推荐，提升了金融产品推荐的时效性。
与现有技术相比，本实施例通过分析挖掘用户最新预设时间段的社交数据，确定该用户的最新金融需求主题，并基于该用户的最新金融需求主题对该用户进行相应金融产品的推荐。由于引入实时动态的社交数据，来预测用户对金融产品的需求偏好，并根据用户对金融产品的需求偏好进行金融产品推荐，能挖掘出用户本身通过社交数据所表达的对金融产品的需求偏好，对于用户金融需求的分析更加精准，能极大地提高金融产品推荐的时效性及效果。
在一可选的实施例中，在上述实施例的基础上，所述步骤S30中根据所述对应关键词所标记的金融需求主题确定用户的最新金融需求主题的步骤包括：
确定预设数量的词汇向量的对应关键词所标记的金融需求主题中各个不同金融需求主题所占的数量；
根据预设数量的词汇向量中各个词汇向量与对应关键词的相似度，以及各个不同金融需求主题所占的数量按预设评分规则计算出各个不同金融需求主题的评分，选择评分最高的金融需求主题作为该用户的最新金融需求主题。
本实施例中，在挑选出与词典中关键词的相似度最高的预设数量如TOP50的词汇向量，并根据挑选出的TOP50的词汇向量的对应关键词所标记的金融需求主题来确定该用户的最新金融需求主题时，可统计挑选出的TOP50的词汇向量的对应关键词所标记的金融需求主题中各个不同金融需求主题的数量或所占比例。进一步地，还可根据TOP50的词汇向量中各个词汇向量与对应关键词的相似度，以及各个不同金融需求主题所占的数量或比例按预设评分规则计算出各个不同金融需求主题的评分，选择评分最高的金融需求主题作为该用户的最新金融需求主题。例如，可预先设定各个不同金融需求主题的数量或所占比例所对应的第一权重值，各个词汇向量与对应关键词的相似度所对应的第二权重值，基于第一权重值、第二权重值综合计算得到各个不同金融需求主题的评分。其中，第一权重值、第二权重值可根据实际应用中是侧重于金融需求主题的占比或侧重于词汇向量与对应关键词的相似度来分别进行设定，能更加准确地确定出用户的最新金融需求主题。
此外，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存储有基于社交数据的金融产品推荐系统，所述基于社交数 据的金融产品推荐系统可被至少一个处理器执行，以使所述至少一个处理器执行如上述实施例中的基于社交数据的金融产品推荐方法的步骤，该基于社交数据的金融产品推荐方法的步骤S10、S20、S30等具体实施过程如上文所述，在此不再赘述。
需要说明的是，在本文中，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、物品或者装置不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、物品或者装置所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括该要素的过程、方法、物品或者装置中还存在另外的相同要素。
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来实现，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。
以上参照附图说明了本申请的优选实施例，并非因此局限本申请的权利范围。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。另外，虽然在流程图中示出了逻辑顺序，但是在某些情况下，可以以不同于此处的顺序执行所示出或描述的步骤。
本领域技术人员不脱离本申请的范围和实质，可以有多种变型方案实现本申请，比如作为一个实施例的特征可用于另一实施例而得到又一实施例。凡在运用本申请的技术构思之内所作的任何修改、等同替换和改进，均应在本申请的权利范围之内。

Claims (20)

  1. 一种电子装置，其特征在于，所述电子装置包括存储器、处理器，所述存储器上存储有可在所述处理器上运行的基于社交数据的金融产品推荐系统，所述基于社交数据的金融产品推荐系统被所述处理器执行时实现如下步骤：
    A1、从预设的数据源获取用户最新预设时间段的社交数据；
    B1、对所述社交数据进行分词，并将各个分词计算转换为词汇向量；
    C1、将每一词汇向量与预设的关键词分别进行相似度计算，确定与每一词汇向量相似度最高的关键词为所述词汇向量的对应关键词，所述关键词预先标记有不同金融需求主题；
    D1、根据对应关键词相似度将所有词汇向量按所述相似度由高到低进行排序，挑选出排序靠前的预设数量的词汇向量；
    E1、获取所述预设数量的词汇向量的对应关键词，并根据所述对应关键词所标记的金融需求主题确定用户的最新金融需求主题；
    F1、基于该用户的最新金融需求主题对该用户进行相应金融产品的推荐。
  2. 如权利要求1所述的电子装置，其特征在于，所述根据所述对应关键词所标记的金融需求主题确定用户的最新金融需求主题包括：
    确定预设数量的词汇向量的对应关键词所标记的金融需求主题中各个不同金融需求主题所占的数量；
    根据预设数量的词汇向量中各个词汇向量与对应关键词的相似度，以及各个不同金融需求主题所占的数量按预设评分规则计算出各个不同金融需求主题的评分，选择评分最高的金融需求主题作为该用户的最新金融需求主题。
  3. 如权利要求1所述的电子装置，其特征在于，所述基于社交数据的金融产品推荐系统被所述处理器执行实现所述步骤B1时，包括：
    利用预设的分词方式对所述社交数据进行分词，并采用词向量模型word2vec将各个分词计算转换为词汇向量。
  4. 如权利要求2所述的电子装置，其特征在于，所述基于社交数据的金融产品推荐系统被所述处理器执行实现所述步骤B1时，包括：
    利用预设的分词方式对所述社交数据进行分词，并采用词向量模型word2vec将各个分词计算转换为词汇向量。
  5. 如权利要求1所述的电子装置，其特征在于，所述基于社交数据的金融产品推荐系统被所述处理器执行实现所述步骤D1时，包括：
    在预设的推荐给该用户的若干金融产品中，将符合该用户的最新金融需求主题的金融产品的推荐优先级提高，以向该用户优先推荐符合该用户的最新金融需求主题的金融产品。
  6. 如权利要求2所述的电子装置，其特征在于，所述基于社交数据的金融产品推荐系统被所述处理器执行实现所述步骤D1时，包括：
    在预设的推荐给该用户的若干金融产品中，将符合该用户的最新金融需求主题的金融产品的推荐优先级提高，以向该用户优先推荐符合该用户的最新金融需求主题的金融产品。
  7. 如权利要求1所述的基于社交数据的金融产品推荐方法，其特征在于，标记的不同金融需求主题包括投资、理财、保险、健康、贷款金融主题。
  8. 一种基于社交数据的金融产品推荐方法，其特征在于，所述基于社交数据的金融产品推荐方法包括：
    A2、从预设的数据源获取用户最新预设时间段的社交数据；
    B2、对所述社交数据进行分词，并将各个分词计算转换为词汇向量；
    C2、将每一词汇向量与预设的关键词分别进行相似度计算，确定与每一词汇向量相似度最高的关键词为所述词汇向量的对应关键词，所述关键词预先标记有不同金融需求主题；
    D2、根据对应关键词相似度将所有词汇向量按所述相似度由高到低进行排序，挑选出排序靠前的预设数量的词汇向量；
    E2、获取所述预设数量的词汇向量的对应关键词，并根据所述对应关键词所标记的金融需求主题确定用户的最新金融需求主题；
    F2、基于该用户的最新金融需求主题对该用户进行相应金融产品的推荐。
  9. 如权利要求8所述的基于社交数据的金融产品推荐方法，其特征在于，所述根据所述对应关键词所标记的金融需求主题确定用户的最新金融需求主题包括：
    确定预设数量的词汇向量的对应关键词所标记的金融需求主题中各个不同金融需求主题所占的数量；
    根据预设数量的词汇向量中各个词汇向量与对应关键词的相似度，以及各个不同金融需求主题所占的数量按预设评分规则计算出各个不同金融需求主题的评分，选择评分最高的金融需求主题作为该用户的最新金融需求主题。
  10. 如权利要求8所述的基于社交数据的金融产品推荐方法，其特征在于，所述步骤B2包括：
    利用预设的分词方式对所述社交数据进行分词，并采用词向量模型word2vec将各个分词计算转换为词汇向量。
  11. 如权利要求9所述的基于社交数据的金融产品推荐方法，其特征在于，所述步骤B2包括：
    利用预设的分词方式对所述社交数据进行分词，并采用词向量模型word2vec将各个分词计算转换为词汇向量。
  12. 如权利要求8所述的基于社交数据的金融产品推荐方法，其特征在于，所述步骤D2包括：
    在预设的推荐给该用户的若干金融产品中，将符合该用户的最新金融需求主题的金融产品的推荐优先级提高，以向该用户优先推荐符合该用户的最新金融需求主题的金融产品。
  13. 如权利要求9所述的基于社交数据的金融产品推荐方法，其特征在于，所述步骤D2包括：
    在预设的推荐给该用户的若干金融产品中，将符合该用户的最新金融需求主题的金融产品的推荐优先级提高，以向该用户优先推荐符合该用户的最新金融需求主题的金融产品。
  14. 如权利要求8所述的基于社交数据的金融产品推荐方法，其特征在于，标记的不同金融需求主题包括投资、理财、保险、健康、贷款金融主题。
  15. 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基于社交数据的金融产品推荐系统，所述基于社交数据的金融产品推荐系统可被至少一个处理器执行，以使所述至少一个处理器执行如下步骤：
    A3、从预设的数据源获取用户最新预设时间段的社交数据；
    B3、对所述社交数据进行分词，并将各个分词计算转换为词汇向 量；
    C3、将每一词汇向量与预设的关键词分别进行相似度计算，确定与每一词汇向量相似度最高的关键词为所述词汇向量的对应关键词，所述关键词预先标记有不同金融需求主题；
    D3、根据对应关键词相似度将所有词汇向量按所述相似度由高到低进行排序，挑选出排序靠前的预设数量的词汇向量；
    E3、获取所述预设数量的词汇向量的对应关键词，并根据所述对应关键词所标记的金融需求主题确定用户的最新金融需求主题；
    F3、基于该用户的最新金融需求主题对该用户进行相应金融产品的推荐。
  16. 如权利要求15所述的计算机可读存储介质，其特征在于，所述根据所述对应关键词所标记的金融需求主题确定用户的最新金融需求主题包括：
    确定预设数量的词汇向量的对应关键词所标记的金融需求主题中各个不同金融需求主题所占的数量；
    根据预设数量的词汇向量中各个词汇向量与对应关键词的相似度，以及各个不同金融需求主题所占的数量按预设评分规则计算出各个不同金融需求主题的评分，选择评分最高的金融需求主题作为该用户的最新金融需求主题。
  17. 如权利要求15所述的计算机可读存储介质，其特征在于，所述基于社交数据的金融产品推荐系统被所述处理器执行实现所述步骤B3时，包括：
    利用预设的分词方式对所述社交数据进行分词，并采用词向量模型word2vec将各个分词计算转换为词汇向量。
  18. 如权利要求16所述的计算机可读存储介质，其特征在于，所述基于社交数据的金融产品推荐系统被所述处理器执行实现所述步骤B3时，包括：
    利用预设的分词方式对所述社交数据进行分词，并采用词向量模型word2vec将各个分词计算转换为词汇向量。
  19. 如权利要求15所述的计算机可读存储介质，其特征在于，所述基于社交数据的金融产品推荐系统被所述处理器执行实现所述步骤D3时，包括：
    在预设的推荐给该用户的若干金融产品中，将符合该用户的最新金融需求主题的金融产品的推荐优先级提高，以向该用户优先推荐符 合该用户的最新金融需求主题的金融产品。
  20. 如权利要求16所述的计算机可读存储介质，其特征在于，所述基于社交数据的金融产品推荐系统被所述处理器执行实现所述步骤D3时，包括：
    在预设的推荐给该用户的若干金融产品中，将符合该用户的最新金融需求主题的金融产品的推荐优先级提高，以向该用户优先推荐符合该用户的最新金融需求主题的金融产品。
PCT/CN2017/108796 2017-08-31 2017-10-31 基于社交数据的金融产品推荐方法、电子装置及介质 WO2019041520A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201710773595.8A CN107704512B (zh) 2017-08-31 2017-08-31 基于社交数据的金融产品推荐方法、电子装置及介质
CN2017107735958 2017-08-31

Publications (1)

Publication Number Publication Date
WO2019041520A1 true WO2019041520A1 (zh) 2019-03-07

Family

ID=61171532

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/108796 WO2019041520A1 (zh) 2017-08-31 2017-10-31 基于社交数据的金融产品推荐方法、电子装置及介质

Country Status (2)

Country Link
CN (1) CN107704512B (zh)
WO (1) WO2019041520A1 (zh)

Families Citing this family (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108765051B (zh) * 2018-04-16 2022-03-22 达而观信息科技（上海）有限公司 物品推荐方法和装置
CN108629693A (zh) * 2018-05-08 2018-10-09 平安科技（深圳）有限公司 自动生成投资建议的方法、装置、计算机设备和存储介质
CN109740914A (zh) * 2018-12-28 2019-05-10 武汉金融资产交易所有限公司 一种金融业务评估、推荐的方法、存储介质、设备及系统
CN110033382B (zh) * 2019-02-12 2020-09-04 阿里巴巴集团控股有限公司 一种保险业务的处理方法、装置及设备
CN109885770B (zh) * 2019-02-20 2022-01-07 杭州威佩网络科技有限公司 一种信息推荐方法、装置、电子设备及存储介质
CN112148958A (zh) * 2019-06-26 2020-12-29 傲基科技股份有限公司 用于信息推荐的方法、设备和计算机存储介质
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
CN111292171B (zh) * 2020-02-28 2023-06-27 中国工商银行股份有限公司 金融理财产品推送方法及装置
CN111967302B (zh) * 2020-06-30 2023-07-25 北京百度网讯科技有限公司 视频标签的生成方法、装置及电子设备
CN113220999A (zh) * 2021-05-14 2021-08-06 北京百度网讯科技有限公司 用户特征的生成方法、装置、电子设备和存储介质
CN114169945B (zh) * 2022-02-08 2022-04-22 北京金堤科技有限公司 一种确定对象所在领域内热门供需产品的方法和装置
CN115577167B (zh) * 2022-08-29 2023-11-21 建信金融科技有限责任公司 基于Webassembly的内容推荐方法、装置、设备及存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140279505A1 (en) * 2013-03-14 2014-09-18 Bank Of America Corporation Recommending vehicle for payment based on social network data
CN105808648A (zh) * 2016-02-25 2016-07-27 焦点科技股份有限公司 一种基于r语言程序的个性化推荐方法
CN106022522A (zh) * 2016-05-20 2016-10-12 南京大学 一种基于互联网公开的大数据预测股票的方法及系统
CN106204156A (zh) * 2016-07-20 2016-12-07 天涯社区网络科技股份有限公司 一种用于网络论坛的广告投放方法及装置
CN106600369A (zh) * 2016-12-09 2017-04-26 广东奡风科技股份有限公司 基于朴素贝叶斯分类的银行金融产品实时推荐系统和方法

Family Cites Families (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9256670B2 (en) * 2013-10-10 2016-02-09 International Business Machines Corporation Visualizing conflicts in online messages
CN104090890B (zh) * 2013-12-12 2016-05-04 深圳市腾讯计算机系统有限公司 关键词相似度获取方法、装置及服务器
CN106156204B (zh) * 2015-04-23 2020-05-29 深圳市腾讯计算机系统有限公司 文本标签的提取方法和装置
CN105488154A (zh) * 2015-11-28 2016-04-13 小米科技有限责任公司 主题应用推荐方法及装置
CN105893611B (zh) * 2016-04-27 2020-04-07 南京邮电大学 一种构建面向社交网络的兴趣主题语义网络的方法
CN106250526A (zh) * 2016-08-05 2016-12-21 浪潮电子信息产业股份有限公司 一种基于内容及用户行为的文本类推荐方法和装置
CN106202574A (zh) * 2016-08-19 2016-12-07 清华大学 面向微博平台话题推荐的评估方法及装置
CN106776881A (zh) * 2016-11-28 2017-05-31 中国科学院软件研究所 一种基于微博平台的领域信息推荐系统及方法
CN107103049A (zh) * 2017-03-31 2017-08-29 努比亚技术有限公司 一种推荐方法及网络设备

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140279505A1 (en) * 2013-03-14 2014-09-18 Bank Of America Corporation Recommending vehicle for payment based on social network data
CN105808648A (zh) * 2016-02-25 2016-07-27 焦点科技股份有限公司 一种基于r语言程序的个性化推荐方法
CN106022522A (zh) * 2016-05-20 2016-10-12 南京大学 一种基于互联网公开的大数据预测股票的方法及系统
CN106204156A (zh) * 2016-07-20 2016-12-07 天涯社区网络科技股份有限公司 一种用于网络论坛的广告投放方法及装置
CN106600369A (zh) * 2016-12-09 2017-04-26 广东奡风科技股份有限公司 基于朴素贝叶斯分类的银行金融产品实时推荐系统和方法

Also Published As

Publication number Publication date
CN107704512B (zh) 2021-08-24
CN107704512A (zh) 2018-02-16

Similar Documents

Publication Publication Date Title
WO2019041520A1 (zh) 基于社交数据的金融产品推荐方法、电子装置及介质
WO2022141861A1 (zh) 情感分类方法、装置、电子设备及存储介质
CN110263248B (zh) 一种信息推送方法、装置、存储介质和服务器
WO2019184217A1 (zh) 热点事件分类方法、装置及存储介质
US8676730B2 (en) Sentiment classifiers based on feature extraction
TWI557664B (zh) Product information publishing method and device
US20180373788A1 (en) Contrastive multilingual business intelligence
US10102191B2 (en) Propagation of changes in master content to variant content
US9767144B2 (en) Search system with query refinement
WO2019061989A1 (zh) 贷款风险控制方法、电子装置及可读存储介质
JP6056610B2 (ja) テキスト情報処理装置、テキスト情報処理方法、及びテキスト情報処理プログラム
WO2020233344A1 (zh) 一种搜索方法、装置及存储介质
US20180181544A1 (en) Systems for Automatically Extracting Job Skills from an Electronic Document
CN107767273B (zh) 基于社交数据的资产配置方法、电子装置及介质
US20160140389A1 (en) Information extraction supporting apparatus and method
US11599586B2 (en) Method and system for providing alternative result for an online search previously with no result
US9772991B2 (en) Text extraction
US11436241B2 (en) Entity resolution based on character string frequency analysis
JPWO2012096388A1 (ja) 意外性判定システム、意外性判定方法およびプログラム
WO2022222942A1 (zh) 问答记录生成方法、装置、电子设备及存储介质
CN106933878B (zh) 一种信息处理方法及装置
CN114387061A (zh) 产品推送方法、装置、电子设备及可读存储介质
TW201241649A (en) Advertisement management device, advertisement selection device, advertisement management method, advertisement management program and storage medium storing advertisement management program
WO2021012958A1 (zh) 原创文本甄别方法、装置、设备与计算机可读存储介质
US20190019094A1 (en) Determining suitability for presentation as a testimonial about an entity

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 PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 30/09/2020)

122 Ep: pct application non-entry in european phase

Ref document number: 17923389

Country of ref document: EP

Kind code of ref document: A1