CN109829108A

CN109829108A - Information recommendation method, device, electronic equipment and readable storage medium storing program for executing

Info

Publication number: CN109829108A
Application number: CN201910081528.9A
Authority: CN
Inventors: 陈文石; 王强; 卢文羊; 李春阳
Original assignee: Beijing Sankuai Online Technology Co Ltd
Current assignee: Beijing Sankuai Online Technology Co Ltd
Priority date: 2019-01-28
Filing date: 2019-01-28
Publication date: 2019-05-31
Anticipated expiration: 2039-01-28
Also published as: CN109829108B; WO2020155877A1

Abstract

The invention discloses a kind of information recommendation method, device, electronic equipment and readable storage medium storing program for executing.The method, comprising: the correlation degree value of the target user Yu each scene theme are determined based on the first user data of target user；According to the corresponding second user data of each information to be recommended, the matching degree value of the information to be recommended Yu the scene theme is determined；Based on the correlation degree value and the matching degree value, the determining and matched information to be recommended of the target user, and the information to be recommended is sent to the target user.Thus existing information recommendation method accuracy and diversity Shortcomings, and the technical problem lower to the attraction of user are solved.Achieve the beneficial effect of the accuracy for improving recommendation information, diversity and user's attraction.

Description

Information recommendation method, device, electronic equipment and readable storage medium storing program for executing

Technical field

The present invention relates to Internet technical fields, and in particular to a kind of information recommendation method, device, electronic equipment and readable Storage medium.

Background technique

With the rise of mobile Internet business, people can be convenient by mobile terminal network visiting, and then obtain Or service needed for customization, therefore O2O (Online To Offline, under line on/line) mode is come into being.This mode Essence be so that user and service each other more convenient and fast discovery, user can select on line at any time oneself needs It is serviced under line；And by the excavation to user's portrait and merchant information, and information recommendation can be carried out to user, and then promoted and used Family experience, and trade company is helped to carry out discovery client.

But the core concept of existing information recommendation scheme is to excavate similar user or similar information to be recommended, is passed through Similar users recommend or recommend similar information to be recommended to user.It can be seen that existing information recommendation method exists more In optimization proposed algorithm, i.e., how user and trade company are more accurately matched, and do not pay close attention to the actually located situation of user, therefore Based on existing information recommended method determine recommendation results accuracy and diversity Shortcomings, and to the attraction of user compared with It is low.

Summary of the invention

The present invention provides a kind of information recommendation method, device, electronic equipment and readable storage medium storing program for executing, partly or entirely to solve The certainly relevant above problem of information recommendation process in the prior art.

According to the present invention in a first aspect, providing a kind of information recommendation method, comprising:

The correlation degree value of the target user Yu each scene theme are determined based on the first user data of target user；

According to the corresponding second user data of each information to be recommended, the information to be recommended and the scene theme are determined Matching degree value；

Based on the correlation degree value and the matching degree value, the determining and matched letter to be recommended of the target user Breath, and the information to be recommended is sent to the target user.

According to the second aspect of the invention, a kind of information recommending apparatus is provided, comprising:

Correlation degree determining module, for the first user data based on target user determine the target user with it is each The correlation degree value of scene theme；

Matching degree determining module, for determining described wait push away according to the corresponding second user data of each information to be recommended Recommend the matching degree value of information Yu the scene theme；

Recommendation information matching module, for being based on the correlation degree value and the matching degree value, it is determining with it is described The matched information to be recommended of target user, and the information to be recommended is sent to the target user.

According to the third aspect of the invention we, a kind of electronic equipment is provided, comprising:

Processor, memory and it is stored in the computer journey that can be run on the memory and on the processor Sequence, which is characterized in that the processor realizes information recommendation method above-mentioned when executing described program.

According to the fourth aspect of the invention, provide a kind of readable storage medium storing program for executing, when the instruction in the storage medium by When the processor of electronic equipment executes, so that electronic equipment is able to carry out information recommendation method above-mentioned.

Information recommendation method according to the present invention can determine that the target is used based on the first user data of target user The correlation degree value at family and each scene theme；According to the corresponding second user data of each information to be recommended, determine described wait push away Recommend the matching degree value of information Yu the scene theme；Based on the correlation degree value and the matching degree value, determine with The matched information to be recommended of target user, and the information to be recommended is sent to the target user.Thus it solves Existing information recommendation method accuracy and diversity Shortcomings, and the technical problem lower to the attraction of user.It obtains Improve the beneficial effect of the accuracy of recommendation information, diversity and user attraction.

The above description is only an overview of the technical scheme of the present invention, in order to better understand the technical means of the present invention, And it can be implemented in accordance with the contents of the specification, and in order to allow above and other objects of the present invention, feature and advantage can It is clearer and more comprehensible, the followings are specific embodiments of the present invention.

Detailed description of the invention

By reading the following detailed description of the preferred embodiment, various other advantages and benefits are common for this field Technical staff will become clear.The drawings are only for the purpose of illustrating a preferred embodiment, and is not considered as to the present invention Limitation.And throughout the drawings, the same reference numbers will be used to refer to the same parts.In the accompanying drawings:

Fig. 1 shows a kind of step flow chart of information recommendation method according to an embodiment of the invention；

Fig. 2 shows a kind of step flow charts of information recommendation method according to an embodiment of the invention；

Fig. 3 shows a kind of structural schematic diagram of information recommending apparatus according to an embodiment of the invention；And

Fig. 4 shows a kind of structural schematic diagram of information recommending apparatus according to an embodiment of the invention.

Specific embodiment

Exemplary embodiments of the present disclosure are described in more detail below with reference to accompanying drawings.Although showing the disclosure in attached drawing Exemplary embodiment, it being understood, however, that may be realized in various forms the disclosure without should be by embodiments set forth here It is limited.On the contrary, these embodiments are provided to facilitate a more thoroughly understanding of the present invention, and can be by the scope of the present disclosure It is fully disclosed to those skilled in the art.

Embodiment one

A kind of information recommendation method provided in an embodiment of the present invention is discussed in detail.

Referring to Fig.1, a kind of step flow chart of information recommendation method in the embodiment of the present invention is shown.

Step 110, the pass of the target user Yu each scene theme are determined based on the first user data of target user Join degree value.

In embodiments of the present invention, when to user's recommendation information, in order to improve feelings locating for recommendation information and relative users The matching degree of scape, to further increase the matching accuracy between recommendation information and relative users.For needing to carry out letter The target user recommended, the first user data of available target user are ceased, and is determined accordingly based on the first user data Target user be directed to each scene theme correlation degree value.

First user data therein may include any available data relevant to target user obtaining.Example Such as, user's representation data, UGC (UserGeneratedContent, user's original content) data, PGC (Professionally-generated Content, professional production content) data, OGC (Occupationally- Generated Content, occupation production content) data, user's location data, current POI (Point ofInterest, Point of interest) data, etc..POI data therein can include but is not limited to POI label, the UGC content of POI, POI association again Article, etc..In embodiments of the present invention, can preset that the first user data included according to demand it is specific in Hold, and in embodiments of the present invention, the first user data can be acquired by any methods availalbe, it is real to this present invention Example is applied to be not limited.

Scene theme therein can also be pre-defined by any methods availalbe according to demand, such as can be passed through The mode that expert defines sets scene theme, data mining can also be carried out by a large amount of reference data, to excavate feelings Scape theme, etc..And the representation of scene theme may include that a scene theme is indicated by least one word, etc. Deng.

For example, can be set, a certain scene theme is the romantic scene theme dated or a certain scene theme is to have a dinner party Scene theme, etc..

Moreover, can determine corresponding mesh based on the first user data using any methods availalbe in embodiments of the present invention User is marked to the correlation degree of each scene theme, this embodiment of the present invention is not limited.For example, can be used based on first Correlation degree value of the user data with the matching degree of each scene theme as corresponding target user and corresponding scene theme, etc. Deng.

Step 120, according to the corresponding second user data of each information to be recommended, the information to be recommended and the feelings are determined The matching degree value of scape theme.

In practical applications, different user can be to a certain extent for the second user data of a certain information to be recommended Characterize corresponding information to be recommended.For example, if containing multiple use in the corresponding second user data of a certain information to be recommended Family selects the corresponding place of the information to be recommended as appointment ground, then can then deduce corresponding information to be recommended and appointment The matching degree of scene theme is higher.

Therefore in embodiments of the present invention, in order to target user recommend with its locating for scene matching degree it is higher to be recommended Information similarly can determine corresponding information to be recommended and each according to the corresponding second user data of each information to be recommended The matching degree value of scene theme.Specifically the corresponding second user number of information to be recommended can be based on by any available means According to determining the matching degree value of the information to be recommended Yu the scene theme, be not limited to this embodiment of the present invention.

Moreover, in embodiments of the present invention, information to be recommended can recommend the information of user, example for any one It such as can include but is not limited to the recommendation information at least one article, the recommendation information at least one place, be directed to Recommendation information of at least one webpage, etc..Specific information to be recommended can be set according to demand, real to this present invention Example is applied to be not limited.

In addition, the corresponding second user data of each recommendation information may include bought for corresponding recommendation information, The second user data of the associated users of operations such as browsing, sharing.And before second user data also can include but is not limited to User's representation data for stating, UGC data, PGC data, OGC data, user's location data, POI data, etc..Specific second The data type that user data is included can be preset according to demand, be not limited to this embodiment of the present invention.

Step 130, it is based on the correlation degree value and the matching degree value, determination is matched with the target user Information to be recommended, and the information to be recommended is sent to the target user.

It has been observed that above-mentioned correlation degree value can characterize the degree of correlation of target user Yu each scene theme, and The degree of correlation of each information to be recommended Yu each scene theme can be then characterized with degree.It is in the purpose of this programme based on mesh It marks scene locating for user and recommends corresponding information to target user.Therefore, then correlation degree value and matching degree can be based on Value, the determining and matched information to be recommended of target user, and then corresponding information to be recommended can be sent to corresponding target User.

Wherein, scene theme corresponding with target user can be determined based on correlation degree value, and is based on matching degree value Can then determine with the matched information to be recommended of corresponding scene theme, and then obtain the matched information to be recommended of target user； Or in embodiments of the present invention, it can also be directly based upon correlation degree value and matching degree value, obtain corresponding information to be recommended With the correlation degree of target user, and then based on correlation degree selected from information to be recommended it is matched to be recommended with target user Information, etc..

Moreover, in embodiments of the present invention, determining information to be recommended can be sent to mesh by any methods availalbe User is marked, this embodiment of the present invention is not limited.

Embodiment two

Referring to Fig. 2, a kind of step flow chart of information recommendation method in the embodiment of the present invention is shown.

Step 210, data mining is carried out based on the third user data that can refer to user, extracts specific scene master Topic.

In practical applications, if being customized by the user scene type of theme, since the demand of different user is different Defined scene theme also can be different, and different scene masters may be set for same scene different user Topic is easy to influence the accuracy of recommendation information to be easy to cause scene theme chaotic.Therefore, in embodiments of the present invention, it is Above situation is avoided, it can the different scene theme of unified definition in advance.It specifically can be based on the third that can refer to user User data carries out data mining, extracts specific scene theme.The third user data that wherein can refer to user specifically may be used To include any available obtain to carry out the user data of scene Topics Crawling.For example, the user on platforms such as purchase by group, take out User data, etc..

And third user data also can include but is not limited to user's representation data above-mentioned, UGC data, PGC data, OGC data, user's location data, POI data, etc..The data type that specific third user data is included can basis Demand is preset, and is not limited to this embodiment of the present invention.

Moreover, in embodiments of the present invention, based on the third user data that can refer to user, any available side can be used Method carries out data mining and is not limited to extract specific scene theme to this embodiment of the present invention.

Optionally, in embodiments of the present invention, the step 210 can further include:

Sub-step 211 carries out vectorization processing to the third user data that can refer to user, obtains the third and use The corresponding multidimensional term vector of user data.

In embodiments of the present invention, scene theme is excavated from the third user data that can refer to user for convenience, Vectorization processing can be carried out to the third user data that can refer to user, to obtain the corresponding multidimensional word of third user data Vector.Specifically vectorization can be carried out to the third user data that can refer to user by any available vector processing method Processing, is not limited this embodiment of the present invention.

For example, vectorization processing can be carried out to the third user data that can refer to user by doc2vec model, or Vectorization processing, etc. can be carried out to the third user data that can refer to user by word2vec model.Moreover, Word2vec model may include skip-gram model, continuous bag of words (continuous bag-of-word, CBOW) mould again Type, etc..

Optionally, in embodiments of the present invention, the sub-step 211, can further include:

Sub-step 2111 carries out word segmentation processing to the third user data that can refer to user；

It in practical applications, then can be to can refer to the of user in order to which third user data is carried out vectorization processing Three user data carry out word segmentation processing, specifically can be using any available participle processing method to the third that can refer to user User data carries out word segmentation processing, is not limited to this embodiment of the present invention.

Sub-step 2112, the invalid word in third user data after removing word segmentation processing, and extract the third user Feature Words in data, the invalid word include at least one of stop words, high frequency words；

In practical applications, the partial words in user data to the determination of scene theme there is no any effect, can be with Defining such word is invalid word, therefore can also not have to the invalid word of consideration when generating multi-C vector, then at through participle Third user data after reason then can further remove invalid word therein, and then extract Feature Words.Invalid word therein It can include but is not limited to stop words (StopWords), high frequency words, etc..

Wherein the particular content of stop words can carry out default setting according to demand, not limited this embodiment of the present invention It is fixed.Such as it is referred to existing such as Harbin Institute of Technology and deactivates deactivated vocabulary setting stop words of vocabulary, the deactivated vocabulary of Baidu, etc.. Moreover, in embodiments of the present invention, it, can according to requirements of service for specific business, special arrangement without help or is not intended to business The word of justice.May include deactivating " sentence " even in stop words, such as electric business " this user does not make comments." can also set It is set to stop words.

It, then can will be remaining after removing invalid word therein for each participle obtained after word segmentation processing Participle is directly as Feature Words；Alternatively, can also further be carried out to obtained each participle after the invalid word of removal further Filtering, and then Feature Words are obtained, specific filtering policy can be preset according to demand, not to this embodiment of the present invention It is limited.

Moreover, word frequency range corresponding to high frequency words can also be preset according to demand, this present invention is implemented Example is also not limited.

Sub-step 2113 constructs the multidimensional term vector of the third user data based on the Feature Words.

After extracting Feature Words, then it can construct to obtain the third user data that can refer to user based on Feature Words Multidimensional term vector.

Different vectorization models is wherein used, the form of obtained multidimensional term vector can be different.For example, multidimensional Term vector can be one-hot coding form or TF-IDF (termfrequency-inverse document Frequency, term frequency-inverse document frequency) form, etc..It can specifically be preset according to demand, to this present invention Embodiment is not limited.

Sub-step 212 is based on the multidimensional term vector, obtains the scene theme by topic model.

After obtaining multidimensional term vector, then it can be based on multidimensional term vector, excavate to obtain scene master by topic model Topic.Topic model therein specifically can be any one available topic model, such as LDA (LatentDirichletAllocation, document subject matter generate model) topic model, Sentence LDA topic model, Copula LDA topic model, etc..

Step 220, according to preset scene Rule of judgment, specific situation theme is defined.

In addition, in embodiments of the present invention, in order to avoid scene type of theme that data mining obtains is not comprehensive enough, or Be it is not accurate enough, specific scene theme can also be defined according to preset scene Rule of judgment.Scene Rule of judgment therein It can then be preset according to demand, this embodiment of the present invention is not limited.Such as, it may be considered that the reality of user Intention and scene, are arranged different scene Rule of judgment, and define the situation of romantic appointment, define situation, etc. of having a dinner party.

It should be noted that in embodiments of the present invention, scene can be acquired based on step 210 and/or step 220 Theme is not limited this embodiment of the present invention.

Optionally, in embodiments of the present invention, the scene theme is by scene descriptor, and/or in the scene theme Theme related term under word class is characterized.

In embodiments of the present invention, for each scene theme of accurate characterization, each scene theme can be set by least One scene descriptor, and/or at least one theme related term under corresponding scene theme word class are characterized.

For example, scene descriptor can be " appointment " for a certain scene theme, and under the scene theme word class Class includes " romance ", " candlelight dinner ", " fresh flower ", etc. theme related term.

Step 230, according to the scene related term for including in every first user data of the target user in each feelings Probability under scape descriptor obtains the characteristic value that first user data is directed to each scene descriptor.

In embodiments of the present invention, by taking LDA model as an example, above-mentioned multidimensional term vector is inputted into LDA model, and then can be with Word-topic matrix is obtained, word therein then can be understood as the theme related term in the embodiment of the present invention, and topic is then It can be understood as the scene descriptor in the present invention.In addition, it is related to obtain each scene based on above-mentioned topic model Probability of the word under corresponding scene descriptor.

So in embodiments of the present invention, in order to obtain target user to the correlation degree value of a certain scene theme, and Since target user may correspond at least one the first user data, and the corresponding different first user data tools of same user The content that body is included can also be different, therefore in embodiments of the present invention, can be with every of target user the first user Data are unit, general under each scene descriptor according to the scene related term for including in every first user data respectively Rate, to obtain the characteristic value that corresponding every first user data is directed to each scene descriptor.

Wherein, a certain first user data is directed to the characteristic value of a certain scene descriptor, with corresponding first user data Corresponding relationship of the scene related term for being included between the probability under corresponding scene descriptor can carry out pre- according to demand If setting, is not limited this embodiment of the present invention.

Optionally, in embodiments of the present invention, the step 230 can further include:

Sub-step 231 extracts the theme related term in first user data for every first user data；

Similarly, since the content for including in the first user data is relatively more, and wherein there may be at least one masters Related term is inscribed, in addition it can include other unrelated words, then in embodiments of the present invention, in order to determine every first number of users According to the characteristic value of corresponding scene descriptor, it can be first directed to every first user data, extract theme related term therein.

Moreover, in embodiments of the present invention, it, can also be right in order to improve the efficiency and accuracy of extracting theme related term Every first user data is pre-processed.Pretreatment therein may include word segmentation processing, the invalid word processing of removal, etc.. Invalid word therein also may include above-mentioned high frequency words and stop words, etc..

Sub-step 232, for each scene descriptor, to probability of the theme related term under the scene descriptor It sums, obtains the characteristic value that first user data is directed to the scene descriptor.

After the scene related term for including in acquiring the first user data, in order to determine that the first user data is corresponding Characteristic value under each scene theme can then be directed to each scene descriptor, mention to from corresponding first user data Probability of the theme related term of taking-up under corresponding scene descriptor is summed, and is obtained corresponding first user data and is directed to The characteristic value of corresponding scene descriptor.

Moreover, in embodiments of the present invention, after extracting scene related term in the first user data, then can obtain Know the corresponding scene descriptor of each scene related term extracted, and then available corresponding first user data is corresponding Scene descriptor.And if not including whole scene related terms under a certain scene descriptor in a certain first user data, that The characteristic value for then illustrating that first user data is directed to the scene descriptor is zero.It therefore, in embodiments of the present invention, can be with Just for the corresponding each scene descriptor of corresponding first user data, to the master of the corresponding scene descriptor of the correspondence extracted It inscribes probability of the related term under corresponding scene descriptor to sum, obtains corresponding first user data for corresponding scene theme The characteristic value of word.

It is assumed that Fuik indicates target user u to the feature for scene descriptor k of the first user data a of information i Value, then characteristic value can be calculated in the following way:

Wherein, n is from the situation related term belonged under scene descriptor k classification extracted in the first user data a Number；f_uiktWhat is indicated is probability of the situation related term t at scene descriptor k, and if not including feelings under scene descriptor k Border related term t, then f_uiktIt can be 0.

Step 240, the characteristic value based on first user data under each scene descriptor obtains the target and uses The correlation degree value for the scene theme that family characterizes the scene descriptor.

In embodiments of the present invention, in every first user data for determining target user under each scene descriptor After characteristic value, then feature of the first user data of target user's whole under each scene descriptor can be based further on Value, obtains corresponding target user to the correlation degree value of each scene theme.

Wherein, the corresponding relationship between characteristic value and correlation degree value can be preset according to demand, to this Inventive embodiments are not limited.

It is every of the target user for example, the correlation degree value that target user is directed to a certain scene theme can be set Characteristic value average value of first user data for the scene descriptor of corresponding scene theme；Or target user's needle can be set To the correlation degree value of a certain scene theme, the feelings of corresponding scene theme are directed to for every first user data of the target user The weighted average of the characteristic value of scape descriptor, weight therein can then carry out presetting, etc. according to demand.

Optionally, in embodiments of the present invention, the step 240 can further include:

Sub-step 241 obtains all the first user data of the target user for institute for each scene descriptor State the characteristic value and value of scene descriptor；

It has been observed that and every first user data of target user can characterize to a certain extent it and be presently in Environment, therefore in embodiments of the present invention, in order to determine that target user, can be with needle currently to the correlation degree of each scene theme To each scene descriptor, all the first user data for obtaining target user respectively are directed to the characteristic value of the scene descriptor And value.

For example, for scene descriptor k, it is assumed that the first user data 1 is n1 to the characteristic value of scene descriptor k, the One user data 2 is n2 to the characteristic value of scene descriptor k, and the first user data 3 is n3 to the characteristic value of scene descriptor k. At this point it is possible to which all the first user data for obtaining target user are directed to the characteristic value of scene descriptor k and value is n1+n2+n3.

It should be noted that the first user data of whole therein can be the whole of available obtained target user First user data is also possible to obtain all the first numbers of users that condition acquires according to preset first user data According to, etc..First user data therein obtains condition and can then be preset according to demand, to this embodiment of the present invention It is not limited.For example, can be set the first user data obtain condition be acquisition issuing time to current time when Between first user data of the difference within the scope of preset time difference, or obtain be directed to presupposed information type the first user data, Etc..

Sub-step 242 obtains the ratio of the quantity of the characteristic value and value and first user data of whole, obtains institute State the correlation degree value for the scene theme that target user is characterized for the scene descriptor.

It acts on to comprehensively consider each first user data to the characterization of scene locating for target user, implements in the present invention Example in, can each first user data for same scene descriptor characteristic value average value as it to corresponding scene theme Correlation degree value, then then available characteristic value and value for a certain scene descriptor is used with the whole first at this time The ratio of the quantity of user data obtains the association journey for the scene theme that the target user is characterized for the scene descriptor Angle value.

For example, all the first user data of target user are directed to scene descriptor k for above-mentioned scene descriptor k Characteristic value and value be n1+n2+n3, and at this time all the first user data specifically include three the first user data, that Available target user is (n1+n2+n3)/3 for the correlation degree of the scene descriptor k scene theme characterized.

Optionally, in embodiments of the present invention, the correlation degree value includes short-term correlation degree value and/or long-term association The weighted sum of degree value；Wherein, the weight of the short-term correlation degree value is greater than the weight of long-term association degree value.

In addition in practical applications, user may be only interested in a certain scene whithin a period of time, it is also possible to long Phase is interested in certain scenes, therefore can be divided into short-term interest and long-term association degree value according to the interest of timeliness user, So corresponding target user also may include short-term correlation degree value and/or length to the correlation degree value of each scene theme The weight of the weighted sum of phase correlation degree value, middle or short term correlation degree value and long-term association degree value can be according to demand It is preset, this embodiment of the present invention is not limited.But general short-term interest can more characterize in practical applications Where the current interest of target user, therefore the weight that the short-term correlation degree value can be set is greater than long-term association degree value Weight, but also can be set short-term correlation degree value weight and long-term association degree value weights sum be 1.

So at this time target user u can be indicated for the correlation degree value of a certain scene theme are as follows:

In formulaIndicate that target user is directed to the short-term correlation degree value of corresponding scene theme,Indicate target user's needle To the long-term association degree value of corresponding scene theme.Wherein α and β is respectively short-term correlation degree value and long-term association degree value Weight.

Wherein, short-term correlation degree value can be the first number of users of target user in the preset one section of short period of consideration According to corresponding correlation degree value, the usually more dynamic interest of short time, and Long-term Interest then can be pre- at least one If interest in longer historical time section, the value mainly determined according to user property and long-term preference, such as user are newborn Youngster mother can have potential interest to parent-offspring's classification for a long time.The type of the corresponding user property of specific Long-term Interest, and it is each The short-term correlation degree value of default shorter historical time section can be preset according to demand, to this embodiment of the present invention It is not limited.

For example, first of target user in the preset time period before short-term correlation degree value is current time can be set Correlation degree value corresponding to user data, can be according to the interest in user's current hour, the same day or other short periods Degree, and Long-term Interest then can be the interest of the primary attribute setting of user, such as young woman for the preference of makeups, or Person's young men for sport and body-building preference etc., or can with active user duplicate click browse chafing dish trade company, I Also may determine that user is potential interested in chafing dish category.Or can also directly be arranged the long-term association degree value be to Short-term correlation degree value in a few default historical time section and value.

Step 250, it for each information to be recommended, obtains and each can refer to every that user is directed to the information to be recommended Second user data.

In embodiments of the present invention, after determining target user to the correlation degree value of each scene theme, in order to Target user recommends applicable information to be recommended, it is also necessary to determine the matching degree of each information to be recommended Yu each scene theme Value.It is specific then can be with reference to second user data corresponding to a certain information to be recommended so that it is determined that the information to be recommended and each The matching degree of a scene theme, therefore firstly, be directed to each information to be recommended, it is available each can refer to user for should Every second user data of information to be recommended.It is therein to can refer to corresponding to user namely above-mentioned whole second user data Can refer to user.

For example, for information i to be recommended, it is assumed that can refer to user u1 for the second user data of information i to be recommended is S1, the second user data that can refer to user u2 for information i to be recommended are s2, then being directed to information i to be recommended, then can be obtained Take above-mentioned second user data s1 and s2.

Step 260, for each scene descriptor, the spy of the scene descriptor is directed to according to the second user data The quantity of value indicative and the corresponding second user data of the information to be recommended obtains the information to be recommended to the scene The matching degree value for the scene theme that descriptor is characterized.

Wherein, matching degree value can carry out pre- according to demand with the corresponding relationship before characteristic value and amount of user data First it is arranged, this embodiment of the present invention is not limited.For example, a certain information to be recommended can be set to a certain scene descriptor The matching degree value of the scene theme characterized is whole second user data corresponding to corresponding information to be recommended for corresponding The characteristic value and value of scene descriptor, the ratio of the quantity of whole second user data corresponding with corresponding information to be recommended.

For example, it is assumed that be directed to scene descriptor k, for information i to be recommended second user data s1 to the scene theme The characteristic value of word is f_uik, then matching degree value of the information i to be recommended to the scene descriptor k scene theme characterized at this time Are as follows:

T_ik=∑_uf_uik/|C_i|

Wherein, ∑_uf_uikIndicate whole second user data corresponding to information i to be recommended for corresponding scene descriptor k Characteristic value and value, | C_i| indicate the quantity of the corresponding whole second user data of information i to be recommended.

In addition, it is necessary to explanation, in embodiments of the present invention, can also predefine each information to be recommended with it is each The matching degree of scene theme, and since second user data can be constantly updated, then can also be using preset time period as the time Interval, periodically obtains currently can refer to the second user data of user and redefine each information to be recommended and each feelings The matching degree of scape theme.

Moreover, if currently having determined that target user to the correlation degree value of each scene theme, may be used also at this time To choose the highest M scene theme of correlation degree value, and then it can only obtain each information to be recommended and the M scene theme Matching degree, without obtaining the matching degree of each information to be recommended and each scene theme.

Step 270, the correlation degree value and the matching degree value are normalized.

In addition, in embodiments of the present invention, for the unified correlation degree to article context pertinent information, can will be associated with Degree value and the matching degree value are normalized.Specific normalization processing method can carry out in advance according to demand Setting, is not limited this embodiment of the present invention.

Step 280, be based on the correlation degree value and the matching degree value, obtain the target user and it is described to Similarity between recommendation information.

Target user is being obtained to the correlation degree value of each scene theme and each information to be recommended and each scene After the matching degree value of theme, then the interested letter to be recommended of target user can be further filtered out from information to be recommended Breath.At this point it is possible to be based further on correlation degree value and the matching degree value, obtain the target user with each wait push away Recommend the similarity between information.Can specifically be determined by any available similarity method obtain target user and it is each to Similarity between recommendation information is not limited this embodiment of the present invention.For example, Euclidean distance (EucledianDistance) similarity, manhatton distance (Manhattan Distance) similarity, Minkowski distance (Minkowskidistance) similarity, cosine similarity (Cosine Similarity), etc..

Optionally, in embodiments of the present invention, the similarity includes cosine similarity.

If similarity is cosine similarity, target user u is with the similarity before information i to be recommended

Cos(I_u,T_i)=| Iu T_i|/||I_u|| ||T_i||

I in formula_uThe level of interest vector of target user, I_u=[I_u1,I_u2,I_u3,...,I_uk], T_iIndicate the matching of article i Degree vector, T_i=[T_i1,T_i2,T_i3,...,T_ik], k ∈ [1, K], wherein I_ukIndicate target user u to k-th scene theme Correlation degree value, T_ikInformation to be recommended is indicated to the matching degree value of k-th of scene theme, K indicates the total quantity of scene theme.

Step 290, it selects with the information to be recommended of the highest preset quantity of target user's similarity as the mesh The recommendation information of user is marked, and the recommendation information is sent to the target user.

After determining the similarity between target user and each information to be recommended, then it can choose and used with the target Recommendation information of the information to be recommended of the highest preset quantity of family similarity as the target user, and by the recommendation information It is sent to the target user.Preset quantity therein can also be preset according to demand, to this embodiment of the present invention It is not limited.

Optionally, in embodiments of the present invention, the user data includes user's original content data, user's representation data At least one of.

Step 2110, according to correlation degree value determination and the matched target context theme of the target user.

Step 2120, according to the target context theme, the information to be recommended is shown.

Optionally, in embodiments of the present invention, the step 2120 can further include: according to the target context master Topic shows that recommendation information relevant to the target context theme in the information to be recommended, the recommendation information include recommending At least one of reason, pictorial information, video information, text information.

In addition, in embodiments of the present invention, can also according to the scene locating for target user, personalization show accordingly to Recommendation information, specifically can be determining with the matched target context theme of target user according to the correlation degree value of aforementioned determination, And then the information to be recommended that can be selected according to target context theme presentation.It specifically can preferentially show information to be recommended Part relevant to target context theme, or if it is wireless network environment that target user, which is presently in scene, at this time Selected information to be recommended can be then shown by the modes such as animation or high definition picture, and if target user is presently in feelings Scape is outdoor environment, then can then show selected recommendation information, etc. by voice mode at this time.

Preferably, recommendation information relevant to the target context theme, institute in selected information to be recommended can be shown State recommendation information can include but is not limited to the matched rationale for the recommendation of the target context theme, pictorial information, video information, At least one of text information.

Moreover, in embodiments of the present invention, it is also based on the third user data progress data mining that can refer to user, Extract specific scene theme；And/or according to preset scene Rule of judgment, define specific situation theme.Also, it is right It is described can refer to user third user data carry out vectorization processing, obtain the corresponding multidimensional word of the third user data to Amount；Based on the multidimensional term vector, the scene theme is obtained by topic model.So as to improve the comprehensive of scene theme Property and accuracy, and then improve recommendation information accuracy and user's attraction.

In addition, in embodiments of the present invention, the scene theme is by scene descriptor, and/or in the scene descriptor Theme related term under classification is characterized.But also can include according in every first user data of the target user Probability of the scene related term under each scene descriptor obtains the spy that first user data is directed to each scene descriptor Value indicative；Characteristic value based on first user data under each scene descriptor obtains the target user to the feelings The correlation degree value for the scene theme that scape descriptor is characterized.In turn, it for every first user data, extracts described first and uses Theme related term in user data；For each scene descriptor, to the theme related term under the scene descriptor Probability is summed, and the characteristic value that first user data is directed to the scene descriptor is obtained.For each scene theme Word, all the first user data for obtaining the target user are directed to the characteristic value and value of the scene descriptor；Described in acquisition The ratio of characteristic value and value and the quantity of first user data of whole obtains the target user for the scene theme The correlation degree value for the scene theme that word is characterized.Also, it is directed to each information to be recommended, acquisition each can refer to user and be directed to Every second user data of the information to be recommended；For each scene descriptor, it is directed to according to the second user data The quantity of the characteristic value of the scene descriptor and the corresponding second user data of the information to be recommended, obtain it is described to The matching degree value for the scene theme that recommendation information characterizes the scene descriptor.Based on the correlation degree value and institute Matching degree value is stated, the similarity between the target user and the information to be recommended is obtained；Selection and the target user Recommendation information of the information to be recommended of the highest preset quantity of similarity as the target user, and the recommendation information is sent out It send to the target user.So as to further increase to obtain the efficiency and accuracy of recommendation information.

It further, in embodiments of the present invention, can also and the target user determining according to the correlation degree value Matched target context theme；According to the target context theme, the information to be recommended is shown.According to the target context master Topic shows recommendation information relevant to the target context theme in the information to be recommended；The recommendation information includes recommending At least one of reason, pictorial information, video information, text information.To realize that treating the personalized of recommendation information shows, Further increase user's attraction of recommendation information.

For embodiment of the method, for simple description, therefore, it is stated as a series of action combinations, but this field Technical staff should be aware of, and embodiment of that present invention are not limited by the describe sequence of actions, because implementing according to the present invention Example, some steps may be performed in other sequences or simultaneously.Secondly, those skilled in the art should also know that, specification Described in embodiment belong to preferred embodiment, the actions involved are not necessarily necessary for embodiments of the present invention.

Embodiment three

A kind of information recommending apparatus provided in an embodiment of the present invention is discussed in detail.

Referring to Fig. 3, a kind of structural schematic diagram of information recommending apparatus in the embodiment of the present invention is shown.

Correlation degree determining module 310, for the first user data based on target user determine the target user with The correlation degree value of each scene theme.

Matching degree determining module 320, for according to the corresponding second user data of each information to be recommended, determine it is described to The matching degree value of recommendation information and the scene theme.

Recommendation information matching module 330, for being based on the correlation degree value and the matching degree value, determining and institute The matched information to be recommended of target user is stated, and the information to be recommended is sent to the target user.

Example IV

Referring to Fig. 4, a kind of structural schematic diagram of information recommending apparatus in the embodiment of the present invention is shown.

Scene Topics Crawling module 410 is extracted for carrying out data mining based on the third user data that can refer to user Specific scene theme out.

Optionally, in embodiments of the present invention, the scene Topics Crawling module 410, can further include:

Vectorization handles submodule, for carrying out vectorization processing to the third user data that can refer to user, obtains To the corresponding multidimensional term vector of the third user data；

Scene Topics Crawling submodule obtains the scene master by topic model for being based on the multidimensional term vector Topic.

Optionally, in embodiments of the present invention, the vectorization handles submodule, comprising:

Word segmentation processing unit, for carrying out word segmentation processing to the third user data that can refer to user；

Feature Words extraction unit, for removing the invalid word in the third user data after word segmentation processing, and described in extracting Feature Words in third user data, the invalid word include at least one of stop words, high frequency words；

Multidimensional term vector construction unit, for constructed based on the Feature Words multidimensional word of the third user data to Amount.

Scene theme definition module 420, for defining specific situation theme according to preset scene Rule of judgment.

Optionally, in embodiments of the present invention, the scene theme is by scene descriptor, and/or in the scene Theme related term under theme word class is characterized.

Correlation degree determining module 430, for the first user data based on target user determine the target user with The correlation degree value of each scene theme.

Wherein, the correlation degree determining module 430, can further include:

Characteristic value acquisition submodule 431, the feelings for including in every first user data according to the target user Probability of the scape related term under each scene descriptor obtains the feature that first user data is directed to each scene descriptor Value；

Correlation degree determines submodule 432, for the spy based on first user data under each scene descriptor Value indicative obtains the correlation degree value for the scene theme that the target user characterizes the scene descriptor.

Optionally, in embodiments of the present invention, the characteristic value acquisition submodule 431, can further include:

Theme related term extraction unit extracts in first user data for being directed to every first user data Theme related term；

Characteristic value acquiring unit, for being directed to each scene descriptor, to the theme related term in the scene theme Probability under word is summed, and the characteristic value that first user data is directed to the scene descriptor is obtained.

Optionally, in embodiments of the present invention, the correlation degree determines submodule 432, can further include:

Characteristic value summation unit obtains all the first users of the target user for being directed to each scene descriptor Data are directed to the characteristic value and value of the scene descriptor；

Correlation degree determination unit, for obtaining the characteristic value and value and the quantity of first user data of whole Ratio obtains the correlation degree value for the scene theme that the target user is characterized for the scene descriptor.

Matching degree determining module 440, for according to the corresponding second user data of each information to be recommended, determine it is described to The matching degree value of recommendation information and the scene theme.

Wherein, in embodiments of the present invention, the matching degree determining module 440, can further include:

User data acquisition submodule 441 obtains for being directed to each information to be recommended and each can refer to user for institute State every second user data of information to be recommended；

Matching degree determines submodule 442, for being directed to each scene descriptor, is directed to according to the second user data The quantity of the characteristic value of the scene descriptor and the corresponding second user data of the information to be recommended, obtain it is described to The matching degree value for the scene theme that recommendation information characterizes the scene descriptor.

Recommendation information matching module 450, for being based on the correlation degree value and the matching degree value, determining and institute The matched information to be recommended of target user is stated, and the information to be recommended is sent to the target user.

Wherein, in embodiments of the present invention, the recommendation information matching module 450, can further include:

Normalized submodule 451, for place to be normalized to the correlation degree value and the matching degree value Reason；

Similarity determines submodule 452, for being based on the correlation degree value and the matching degree value, described in acquisition Similarity between target user and the information to be recommended；

Recommendation information matched sub-block 453, for select with the highest preset quantity of target user's similarity to Recommendation information of the recommendation information as the target user, and the recommendation information is sent to the target user.

Target context theme determining module 460 is matched for being determined according to the correlation degree value with the target user Target context theme.

Recommendation information display module 470, for showing the information to be recommended according to the target context theme.

Optionally, in embodiments of the present invention, the recommendation information display module, can further include:

Recommendation information shows submodule, for according to the target context theme, show in the information to be recommended with institute State the relevant recommendation information of target context theme；The recommendation information includes rationale for the recommendation, pictorial information, video information, text At least one of information.

A kind of electronic equipment is also disclosed in the embodiment of the present invention, comprising:

A kind of readable storage medium storing program for executing is also disclosed in the embodiment of the present invention, when the instruction in the storage medium is set by electronics When standby processor executes, so that electronic equipment is able to carry out information recommendation method above-mentioned.

For device embodiment, since it is basically similar to the method embodiment, related so being described relatively simple Place illustrates referring to the part of embodiment of the method.

Algorithm and display are not inherently related to any particular computer, virtual system, or other device provided herein. Various general-purpose systems can also be used together with teachings based herein.As described above, it constructs required by this kind of system Structure be obvious.In addition, the present invention is also not directed to any particular programming language.It should be understood that can use various Programming language realizes summary of the invention described herein, and the description done above to language-specific is to disclose this hair Bright preferred forms.

In the instructions provided here, numerous specific details are set forth.It is to be appreciated, however, that implementation of the invention Example can be practiced without these specific details.In some instances, well known method, structure is not been shown in detail And technology, so as not to obscure the understanding of this specification.

Similarly, it should be understood that in order to simplify the disclosure and help to understand one or more of the various inventive aspects, Above in the description of exemplary embodiment of the present invention, each feature of the invention is grouped together into single implementation sometimes In example, figure or descriptions thereof.However, the disclosed method should not be interpreted as reflecting the following intention: i.e. required to protect Shield the present invention claims features more more than feature expressly recited in each claim.More precisely, as following Claims reflect as, inventive aspect is all features less than single embodiment disclosed above.Therefore, Thus the claims for following specific embodiment are expressly incorporated in the specific embodiment, wherein each claim itself All as a separate embodiment of the present invention.

Those skilled in the art will understand that can be carried out adaptively to the module in the equipment in embodiment Change and they are arranged in one or more devices different from this embodiment.It can be the module or list in embodiment Member or component are combined into a module or unit or component, and furthermore they can be divided into multiple submodule or subelement or Sub-component.Other than such feature and/or at least some of process or unit exclude each other, it can use any Combination is to all features disclosed in this specification (including adjoint claim, abstract and attached drawing) and so disclosed All process or units of what method or apparatus are combined.Unless expressly stated otherwise, this specification is (including adjoint power Benefit require, abstract and attached drawing) disclosed in each feature can carry out generation with an alternative feature that provides the same, equivalent, or similar purpose It replaces.

In addition, it will be appreciated by those of skill in the art that although some embodiments described herein include other embodiments In included certain features rather than other feature, but the combination of the feature of different embodiments mean it is of the invention Within the scope of and form different embodiments.For example, in the following claims, embodiment claimed is appointed Meaning one of can in any combination mode come using.

Various component embodiments of the invention can be implemented in hardware, or to run on one or more processors Software module realize, or be implemented in a combination thereof.It will be understood by those of skill in the art that can be used in practice Microprocessor or digital signal processor (DSP) come realize some in information recommendation equipment according to an embodiment of the present invention or The some or all functions of person's whole component.The present invention is also implemented as one for executing method as described herein Point or whole device or device programs (for example, computer program and computer program product).Such this hair of realization Bright program can store on a computer-readable medium, or may be in the form of one or more signals.It is such Signal can be downloaded from an internet website to obtain, and is perhaps provided on the carrier signal or is provided in any other form.

It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and ability Field technique personnel can be designed alternative embodiment without departing from the scope of the appended claims.In the claims, Any reference symbol between parentheses should not be configured to limitations on claims.Word "comprising" does not exclude the presence of not Element or step listed in the claims.Word "a" or "an" located in front of the element does not exclude the presence of multiple such Element.The present invention can be by means of including the hardware of several different elements and being come by means of properly programmed computer real It is existing.In the unit claims listing several devices, several in these devices can be through the same hardware branch To embody.The use of word first, second, and third does not indicate any sequence.These words can be explained and be run after fame Claim.

Claims

1. a kind of information recommendation method characterized by comprising

According to the corresponding second user data of each information to be recommended, the matching of the information to be recommended Yu the scene theme is determined Degree value；

Based on the correlation degree value and the matching degree value, the determining and matched information to be recommended of the target user.

2. the method according to claim 1, wherein being determined in first user data based on target user Before the step of correlation degree value of the target user and each scene theme, further includes:

Data mining is carried out based on the third user data that can refer to user, extracts specific scene theme；

And/or according to preset scene Rule of judgment, define specific situation theme.

3. according to the method described in claim 2, it is characterized in that, described carried out based on the third user data that can refer to user Data mining, the step of extracting specific scene theme, comprising:

Vectorization processing is carried out to the third user data that can refer to user, it is corresponding more to obtain the third user data Tie up term vector；

Based on the multidimensional term vector, the scene theme is obtained by topic model.

4. according to the method described in claim 3, it is characterized in that, it is described to the third user data that can refer to user into Row vectorization processing, the step of obtaining the third user data corresponding multidimensional term vector, comprising:

Word segmentation processing is carried out to the third user data that can refer to user；

The invalid word in third user data after removing word segmentation processing, and the Feature Words in the third user data are extracted, The invalid word includes at least one of stop words, high frequency words；

The multidimensional term vector of the third user data is constructed based on the Feature Words.

5. method according to claim 1-3, which is characterized in that the scene theme by scene descriptor, and/ Or the theme related term under the scene theme word class is characterized.

6. according to the method described in claim 5, it is characterized in that, first user data based on target user determines institute The step of stating correlation degree value of the target user with each scene theme, comprising:

According to the scene related term for including in every first user data of the target user under each scene descriptor Probability obtains the characteristic value that first user data is directed to each scene descriptor；

Characteristic value based on first user data under each scene descriptor obtains the target user to the scene The correlation degree value for the scene theme that descriptor is characterized.

7. according to the method described in claim 6, it is characterized in that, every first number of users according to the target user Probability of the scene related term for including under each scene descriptor obtains first user data and is directed to each scene The step of characteristic value of descriptor, comprising:

For every first user data, the theme related term in first user data is extracted；

For each scene descriptor, sums, obtain to probability of the theme related term under the scene descriptor First user data is directed to the characteristic value of the scene descriptor.

8. according to the method described in claim 6, it is characterized in that, described be based on first user data in each scene master Characteristic value under epigraph obtains the correlation degree value for the scene theme that the target user characterizes the scene descriptor Step, comprising:

For each scene descriptor, all the first user data of the target user are obtained for the scene descriptor Characteristic value and value；

The ratio for obtaining the quantity of the characteristic value and value and first user data of whole, obtains the target user and is directed to The correlation degree value for the scene theme that the scene descriptor is characterized.

9. the method according to claim 6 or 8, which is characterized in that the correlation degree value includes short-term correlation degree value And/or the weighted sum of long-term association degree value；Wherein, the weight of the short-term correlation degree value is greater than long-term association degree value Weight.

10. according to the method described in claim 5, it is characterized in that, described according to the corresponding second user of each information to be recommended Data, the step of determining the matching degree value of the information to be recommended and the scene theme, comprising:

For each information to be recommended, obtaining each can refer to every second user number that user is directed to the information to be recommended According to；

For each scene descriptor, the characteristic value of the scene descriptor, Yi Jisuo are directed to according to the second user data The quantity of the corresponding second user data of information to be recommended is stated, the information to be recommended is obtained and the scene descriptor is characterized Scene theme matching degree value.

11. the method according to claim 1, wherein described be based on the correlation degree value and the matching Degree value, the determining and matched information to be recommended of the target user, and the information to be recommended is sent to the target and is used The step of family, comprising:

Based on the correlation degree value and the matching degree value, obtain between the target user and the information to be recommended Similarity；

It selects and recommendation of the information to be recommended of the highest preset quantity of target user's similarity as the target user Information, and the recommendation information is sent to the target user.

12. according to the method for claim 11, being based on the correlation degree value and the matching degree value described, obtain Before the step of taking the similarity between the target user and the information to be recommended, further includes:

The correlation degree value and the matching degree value are normalized.

13. the method according to claim 1, wherein being based on the correlation degree value and described described With degree value, the determining and matched information to be recommended of the target user, and the information to be recommended is sent to the target After the step of user, further includes:

According to correlation degree value determination and the matched target context theme of the target user；

According to the target context theme, the information to be recommended is shown.

14. according to the method for claim 13, which is characterized in that it is described according to the target context theme, described in displaying The step of information to be recommended, comprising:

According to the target context theme, recommendation relevant to the target context theme in the information to be recommended is shown Breath, the recommendation information includes at least one of rationale for the recommendation, pictorial information, video information, text information.

15. a kind of information recommending apparatus characterized by comprising

Correlation degree determining module determines the target user and each scene for the first user data based on target user The correlation degree value of theme；

Matching degree determining module, for determining the letter to be recommended according to the corresponding second user data of each information to be recommended The matching degree value of breath and the scene theme；

Recommendation information matching module, for being based on the correlation degree value and the matching degree value, the determining and target The matched information to be recommended of user, and the information to be recommended is sent to the target user.

16. a kind of electronic equipment characterized by comprising

Processor, memory and it is stored in the computer program that can be run on the memory and on the processor, It is characterized in that, the processor realizes the information as described in any one of claim 1-14 when executing the computer program Recommended method.

17. a kind of readable storage medium storing program for executing, which is characterized in that when the instruction in the storage medium is held by the processor of electronic equipment When row, so that electronic equipment is able to carry out the information recommendation method as described in any one of claim 1-14.