Recommended methods, apparatuses, and media

Abstract

Embodiments of the present application provide a recommendation method, device and medium. The method specifically comprises: determining matching information between candidate content and user portrait features; the user portrait features are obtained according to user usage behavior data for input method provided candidate items and labels of the candidate items; and the candidate content is recommended according to the matching information. Embodiments of the present application can improve the accuracy of recommendation; in the recommendation scene of the input method, the recommended candidate items can meet the personalized input intention of the user.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the technical field of input, in particular to a recommendation method, device and medium. BACKGROUND

[0002] Input method refers to an encoding method adopted to input characters into a computer or other devices (such as mobile phones, tablet computers, etc.), which can be applied to various scenarios. For example, a user can input keywords in a search engine to search web pages, input texts in an instant messaging APP (application) to communicate with other users, and input texts in a document APP to edit documents, etc.

[0003] The association function is an extended function of the input method, which reduces the number of times of active input and keystrokes of the user and increases the intelligence of the input method. Currently, the input method can recommend corresponding association candidates according to the already on-screen content of the user.

[0004] The inventors found in the implementation of the embodiments of the present application that the same association candidate is usually recommended for different users in the case of the same already on-screen content. Different users usually have different input intentions, and therefore the currently recommended association candidate may not meet the personalized input intention of the user. SUMMARY

[0005] Embodiments of the present application provide a recommendation method, device and medium, which can improve the accuracy of recommendation, and the recommended candidate can meet the personalized input intention of the user in the recommendation scenario of the input method.

[0006] To solve the above problem, the embodiments of the present application disclose a recommendation method, comprising:

[0007] determining matching information between the candidate content and user portrait features, wherein the user portrait features are obtained according to the use behavior data of the user for the candidate provided by the input method and the label of the candidate;

[0008] recommending the candidate content according to the matching information.

[0009] To solve the above problem, the embodiments of the present application disclose a recommendation method, comprising:

[0010] adding a label to the candidate provided by the input method;

[0011] determining the use behavior data of the user for the candidate;

[0012] determining the user portrait features corresponding to the user according to the use behavior data and the label of the candidate.

[0013] In another aspect, embodiments of the present application disclose a recommending apparatus, comprising:

[0014] a determining module configured to determine matching information between candidate content and user portrait features, wherein the user portrait features are obtained according to user behavior data of a candidate item provided by an input method and labels of the candidate item; and

[0015] a recommending module configured to recommend the candidate content according to the matching information.

[0016] In another aspect, embodiments of the present application disclose a recommending apparatus, comprising:

[0017] a label adding module configured to add labels to a candidate item provided by an input method;

[0018] a data determining module configured to determine user behavior data of the candidate item; and

[0019] a portrait determining module configured to determine user portrait features of the user according to the user behavior data and the labels of the candidate item.

[0020] In yet another aspect, embodiments of the present application disclose an apparatus for recommending, comprising a memory and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by one or more processors, and the one or more programs contain instructions for performing the following operations:

[0021] determining matching information between candidate content and user portrait features, wherein the user portrait features are obtained according to user behavior data of a candidate item provided by an input method and labels of the candidate item;

[0022] recommending the candidate content according to the matching information.

[0023] In yet another aspect, embodiments of the present application disclose an apparatus for recommending, comprising a memory and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by one or more processors, and the one or more programs contain instructions for performing the following operations:

[0024] adding labels to a candidate item provided by an input method;

[0025] determining user behavior data of the candidate item;

[0026] determining user portrait features of the user according to the user behavior data and the labels of the candidate item.

[0027] In yet another aspect, the embodiments of the present application disclose a machine readable medium having instructions stored thereon that, when executed by one or more processors, cause an apparatus to perform the recommendation method according to one or more of the preceding aspects.

[0028] The embodiments of the present application have the following advantages:

[0029] In the embodiments of the present application, the candidate provided by the input method can be provided with a label, and the embodiments of the present application obtain the user portrait feature according to the use behavior data of the user for the candidate and the label of the candidate. The use behavior data can represent the use preference of the user for the candidate. In this way, the embodiments of the present application can obtain the user portrait feature that can reflect the user preference according to the label of the candidate preferred by the user, and improve the accuracy of the user portrait feature.

[0030] Further, the embodiments of the present application recommend the candidate content according to the user portrait feature with higher accuracy, which can improve the accuracy of the recommendation. In the case where the candidate content is the candidate provided by the input method, since the user portrait feature can reflect the input preference of the user, the candidate recommended by the embodiments of the present application can meet the personalized input intention of the user. BRIEF DESCRIPTION OF DRAWINGS

[0031] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the description of the embodiments of the present application. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0032] Figure 1 is a schematic diagram of an application environment of a recommendation method according to an embodiment of the present application;

[0033] Figure 2 is a step flowchart of a recommendation method according to an embodiment of the present application;

[0034] Figure 3 is a step flowchart of a recommendation method according to another embodiment of the present application;

[0035] Figure 4 is a structural block diagram of a recommendation device according to an embodiment of the present application;

[0036] Figure 5 is a structural block diagram of a recommendation device according to an embodiment of the present application;

[0037] Figure 6 is a block diagram of a device 800 for recommendation according to an embodiment of the present application; and

[0038] Figure 7is a structural schematic diagram of a server in some embodiments of the present application. DETAILED DESCRIPTION

[0039] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative effort should fall within the protection scope of the present application.

[0040] The embodiments of the present application can be applied to a recommendation scenario, which can be used for recommending candidate content. The types of the candidate content can include: candidate content of an input type provided by an input method (hereinafter referred to as a candidate item), candidate content of a commodity type, or candidate content of a news type, and the like. The candidate item of the input type can be used in an input process, and the candidate item of the input type can be displayed on a screen to a corresponding application program, for example, the candidate item of the input type can be displayed on an input box of an instant messaging program or a search program. The candidate content of the commodity type can include: a link of a commodity and the like. The candidate content of the news type can include: a link of a news and the like. It can be understood that the embodiments of the present application do not limit the specific types of the candidate content.

[0041] In view of the technical problem that the input method recommended association candidate may not meet the personalized input intention of the user, the embodiments of the present application provide a recommendation scheme, which can include: determining matching information between candidate content and user portrait features; the user portrait features can be obtained according to user usage behavior data for candidate items provided by an input method and labels of the candidate items; and the candidate content is recommended according to the matching information.

[0042] In the embodiments of the present application, the candidate items provided by the input method can have labels, and the embodiments of the present application obtain user portrait features according to user usage behavior data for the candidate items and the labels of the candidate items. The usage behavior data can represent the user's usage preference for the candidate items. In this way, the embodiments of the present application can obtain user portrait features that can reflect the user's preference according to the labels of the candidate items preferred by the user, thereby improving the accuracy of the user portrait features.

[0043] Further, the embodiments of the present application recommend the candidate content according to the user portrait features with higher accuracy, which can improve the accuracy of the recommendation. In the case where the candidate content is the candidate items provided by the input method, since the user portrait features can reflect the input preference of the user, the candidate items recommended by the embodiments of the present application can meet the personalized input intention of the user.

[0044] For example, in the recommendation scenario of the input method, the candidate items corresponding to the already on-screen content "I want to buy" specifically include "clothes", "this", "things", "lipstick", "game console", etc. In the traditional technology, the same candidate items are recommended to different users.

[0045] However, the embodiment of the present application can obtain the user portrait feature according to the user's usage behavior data for the above candidate items and the labels of the above candidate items. For example, if user A uses the candidate item with the label "female" multiple times, it can be considered that the user portrait feature of user A includes "female", and therefore, in the case of the already on-screen content "I want to buy" of user A, candidate items that meet "female" can be recommended, such as "clothes", "lipstick", etc. Further, if user A uses the candidate item with the label "makeup" multiple times, it can be considered that the user portrait feature of user A also includes "makeup", and therefore, candidate items that meet "female" and "makeup" can be recommended, such as "lipstick", etc.

[0046] For another example, if user B uses the candidate item with the label "male" and "electronic product" multiple times, it can be considered that the user portrait feature of user B includes "male" and "electronic product", and therefore, in the case of the already on-screen content "I want to buy" of user B, candidate items that meet "male" and "electronic product" can be recommended, such as "game console", etc.

[0047] The recommendation method provided by the embodiment of the present application can be applied to Figure 1 as shown in the application environment Figure 1 The client 100 and the server 200 are located in a wired or wireless network, and the client 100 and the server 200 perform data interaction through the wired or wireless network.

[0048] Optionally, the client 100 can run on a terminal, and the terminal specifically includes but is not limited to: a smart phone, a tablet computer, an e-book reader, an MP3 (Moving Picture Experts Group Audio Layer III) player, an MP4 (Moving Picture Experts Group Audio Layer IV) player, a laptop computer, a vehicle-mounted computer, a desktop computer, a set-top box, a smart television, a wearable device, etc. The client 100 can correspond to a website or an APP (Application). The client 100 can correspond to an input method APP, an instant messaging APP, etc.

[0049] The service end of the embodiment of the application can be a cloud service end. The cloud service end is a kind of computing service with simple and efficient, safe and reliable, and elastic processing capability. The resource information of the cloud service end is dynamic, so that the processing capability thereof is elastic.

[0050] The embodiment of the application can be applied to an input method program of various input modes such as keyboard symbols, handwriting, and voice. Taking the keyboard symbol input mode as an example, a user can input characters by using an encoded string, and the input string can refer to the encoded string input by the user. In the input method field, for an input method program of Chinese, Japanese, Korean, or other languages, the input string input by the user can generally be converted into a candidate of the corresponding language. Hereinafter, Chinese is mainly taken as an example for description, and other languages such as Japanese and Korean can be referred to each other. It can be understood that the Chinese input method described above can include, but is not limited to, full spelling, simplified spelling, strokes, and five-stroke input, and the embodiment of the application does not limit the specific input method program corresponding to a certain language.

[0051] Taking the input of Chinese as an example, the type of the encoded string can include a pinyin string and a character string (such as a five-stroke string). Taking the input of English as an example, the type of the encoded string can include a letter string.

[0052] In actual application, for the input mode of keyboard symbols, the user can input the input string by using a physical keyboard or a virtual keyboard. For example, for a terminal with a touch screen, a virtual keyboard can be set in an input interface to input the input string by triggering a virtual key included in the virtual keyboard. Optionally, examples of the virtual keyboard can include a 9-key keyboard and a 26-key keyboard. It can be understood that, in addition to the virtual keys corresponding to letters, the input interface can also be provided with symbol keys, number keys, function keys such as Chinese-English switching keys, or tool bar keys, and it can be understood that the embodiment of the application does not limit the specific keys included in the input interface.

[0053] According to some embodiments, the input string can include, but is not limited to, one key symbol or a combination of multiple key symbols input by the user. The key symbol can specifically include pinyin, strokes, and hiragana.

[0054] In the embodiment of the application, the candidate can be used to represent one or more characters provided by the input method program and to be selected by the user. The candidate can be a character of Chinese, English, Japanese, or other languages, and the candidate can also be a combination of symbols in the form of color text, expression, and picture. The color text includes, but is not limited to, a picture composed of lines, symbols, and characters, for example, examples of the color text can include “: P”, “:-o”, and “:-)”.

[0055] Method Example 1

[0056] Reference Figure 2 The diagram illustrates a flowchart of a recommended method embodiment of the present invention, which may specifically include the following steps:

[0057] Step 201: Add tags to the candidate input methods;

[0058] Step 202: Determine the user's usage behavior data for the above candidates;

[0059] Step 203: Based on the above usage behavior data and the tags of the above candidate options, determine the user profile characteristics corresponding to the above users.

[0060] Figure 2 The method embodiment shown is used to determine the user profile features corresponding to the user based on the user's usage behavior data of the candidate options and the tags of the candidate options. Figure 2 The method embodiments shown can be executed by the client and / or the server. It is understood that the embodiments of the present invention do not limit the specific subject of execution of the method embodiments.

[0061] In step 201, the candidate options provided by the input method may include: character candidate options, or emoji candidate options, etc. When the candidate option is a character, the corresponding language unit may include: character, word, phrase, or sentence, etc.

[0062] This invention allows for pre-adding tags to entries in a source database for candidate options. The invention also allows for searching the source database based on context and / or the input string to obtain candidate options. The source database for candidate options may include a thesaurus or an emoji database, etc. The thesaurus may include a system thesaurus, a user thesaurus, a cell thesaurus, a cloud thesaurus, etc. For example, tags can be pre-added to entries in the thesaurus, or tags can be pre-added to emojis in the emoji database, etc. Context can represent the context corresponding to the input cursor, and the context may include: content already displayed by the local user, and / or, communication content sent by the peer user.

[0063] In this embodiment of the invention, the purpose of a user using candidate options is usually to express themselves. Therefore, the labels on candidate options can characterize the features of the user who uses the candidate options to express themselves. For example, a user who uses the candidate option "husband" to express themselves usually has the characteristic of being "female," so the candidate option "husband" can be labeled "female."

[0064] In an optional embodiment of the present application, the above-mentioned label can comprise a language style label. The language style label can be used to represent the style of the language used by the candidate. Examples of the language style can include: cute style, lively style, serious style, sweet style, two-dimensional style, counterattack style, straight man style, bookish style, etc.

[0065] In another optional embodiment of the present application, the candidate is applied to a communication environment, and the label can comprise at least one of the following features: a first user label of a local user, a second user label of at least one opposite user, and a user relationship label of multi-end users.

[0066] The local user can communicate with at least one opposite user, and therefore, the multi-end users in the communication environment can comprise two users or more than two users. For example, in the communication environment of a group chat, the multi-end users can comprise more than two users.

[0067] The local user can represent a user who uses the candidate to express. The opposite user can represent a user on the opposite end of the communication. The user relationship can represent the relationship between the local user and the opposite user.

[0068] For example, a user who uses the candidate X "Husband, what are you doing" to express usually has the "female" feature, and the candidate X can also represent the "male" feature of the opposite user and the user relationship "husband and wife or lovers". Therefore, the embodiments of the present application can add the following labels to the candidate X "Husband, what are you doing": the first user label

female

male

lovers

[0069] In the embodiments of the present application, the labels can be added to the candidates provided by the input method in an artificial manner or a machine manner.

[0070] In an optional embodiment of the present application, the adding manner for adding the labels to the candidates provided by the input method can specifically comprise:

[0071] Adding manner 1, determining the label corresponding to the candidate according to semantic information corresponding to the candidate; and / or

[0072] Adding manner 2, determining the label corresponding to the candidate according to a classification model corresponding to the label; and / or

[0073] Adding manner 3, determining the label corresponding to the candidate according to a mapping relationship between the candidate and the label.

[0074] For adding manner 1, the semantic information can contain or imply the information of the first user label, the second user label, and the user relationship label, and therefore, the embodiments of the present application can determine the label corresponding to the candidate according to the semantic information.

[0075] The embodiment of the present application can determine semantic information by using a natural language understanding method. The natural language understanding method can include a keyword extraction method, a syntax analysis method, a machine learning method, etc. It can be understood that the embodiment of the present application does not limit the specific natural language understanding method.

[0076] In actual application, a keyword can be extracted from the candidate, and the label corresponding to the candidate can be determined according to the keyword. For example, the keyword "husband" can be extracted from the candidate X "Husband, what are you doing". Since the user who uses the candidate "husband" to express usually has the "female" feature, the "female" label can be added to the candidate X "Husband, what are you doing". In addition, the chat object corresponding to the candidate X can be determined to be "male" and the user relationship to be "lover" based on natural language understanding, so that the corresponding label can be obtained.

[0077] For the adding method 2, the classification model can have the classification ability of the label corresponding category. The classification model can be a two-class classification model or a multi-class classification model.

[0078] The two-class classification model is used to identify whether the candidate belongs to the target category or the non-target category.

[0079] The multi-class classification model is used to identify which one of the multiple target categories the candidate belongs to. The multiple target categories corresponding to the multi-class classification model can include categories corresponding to multiple labels, such as the "female" category, the "male" category, the "cute style" category, the "lively style" category, the "serious style" category, the "sweet style" category, etc.

[0080] In an optional embodiment of the present application, the mathematical model can be trained based on the training data to obtain the classification model, which can represent the mapping relationship between the input data (candidate) and the output data (label of the candidate). The output data of the classification model can include the label of the candidate or the probability that the candidate belongs to a certain label.

[0081] The training data corresponding to the classification model can include the candidate of the label corresponding category. Assuming that the label corresponding category is "female", the training data can include the candidate corresponding to "female". Training the mathematical model based on the candidate of the label corresponding category can make the obtained classification model have the identification ability of the corresponding category, that is, the obtained classification model can identify whether the input candidate belongs to the corresponding category.

[0082] A mathematical model is a scientific or engineering model constructed by using mathematical logic and mathematical language. The mathematical model is a mathematical structure that is a relationship structure depicted by mathematical symbols, which is used to describe the characteristics or quantitative dependence relationship of a certain thing system. The mathematical model can be one or a group of algebraic equations, differential equations, difference equations, integral equations or statistical equations and their combinations, which quantitatively or qualitatively describe the mutual relationship or causal relationship between variables of the system. In addition to the mathematical model described by equations, there are models described by other mathematical tools such as algebra, geometry, topology, mathematical logic, etc. Among them, the mathematical model describes the behavior and characteristics of the system rather than the actual structure of the system. The mathematical model can be trained by using machine learning, deep learning methods, etc. The machine learning method can include linear regression, decision tree, random forest, etc. The deep learning method can include convolutional neural network (CNN), long short-term memory (LSTM), gated recurrent unit (GRU), etc.

[0083] For the adding method 3, the candidate can be looked up in the mapping relationship between the candidate and the label according to the candidate to obtain the label corresponding to the candidate.

[0084] In an optional embodiment of the present application, the corresponding candidate can be constructed for the preset label, and the mapping relationship between the candidate and the label is established. The above-mentioned construction of the corresponding candidate for the preset label can expand the range of the mapping relationship.

[0085] Optionally, the construction of the corresponding candidate can include the following construction methods:

[0086] Construction method 1: constructing the corresponding candidate according to the preset label and the preset context; and / or

[0087] Construction method 2: constructing the second candidate corresponding to the second language style label according to the first language style label and the first candidate corresponding to the first language style label, and the second language style label.

[0088] The construction method 1 can construct the corresponding candidate for a specific preset label and a specific preset context. For example, in the case of the preset context being "tomorrow Saturday", the candidate "not going to school" is constructed for the preset label "student", and the candidate "not going to work" is constructed for the preset label "employee".

[0089] Construction Method 2 utilizes style transfer to construct second candidate options for the second language style tags, given the first language style tag and its corresponding first candidate, as well as the second language style tag. The first and second candidate options can correspond to the same or similar expressive intent, but with different language styles. For example, for the expressive intent "comfort," the first candidate option corresponding to the "serious style" tag could be "comfort," while the second candidate option corresponding to the "cute style" tag could be "mwah."

[0090] When constructing second candidate options for second language style tags, a preset context can also be considered. For example, if the preset context is "I'm going to sleep," then under the "lively style" tag, the corresponding candidate option could be "Don't sleep, get up and have fun"; under the "serious style" tag, the corresponding candidate option could be "Good night."

[0091] In an optional embodiment of this application, a mathematical model can be trained based on training data to obtain a style transfer model. The style transfer model can represent the mapping relationship between input data (a first language style tag and its corresponding first candidate, and a second language style tag) and output data (a second candidate). The style transfer model can be used to transfer the expression corresponding to the first candidate from the first language style tag to the second language style tag.

[0092] The training data for a style transfer model can include: first language style tags and their corresponding first candidate samples, and second language style tags and their corresponding second candidate samples. The style transfer model can acquire style transfer capabilities based on the training data.

[0093] In step 202, the input method program can provide the user with candidate options during the input process for the user to choose from. The user's behavior data regarding these candidate options can include the data on which the user selects and displays the candidate option. For example, if the input method program provides candidate option A and candidate option B, and the user selects to display candidate option B, then the user's behavior data regarding candidate option B can be recorded.

[0094] It is understandable that, in addition to the data displayed on the screen, the user behavior data for the above-mentioned candidate options may also include: sending data after displaying the screen, searching data after displaying the screen, or sharing data after displaying the screen, browsing data of the content corresponding to the candidate options (such as videos, microblogs, novels, etc.).

[0095] The aforementioned usage behavior data may include information such as candidate options, usage behavior, and usage time. Alternatively, the aforementioned usage behavior data may include information such as candidate options, usage behavior, and number of uses.

[0096] In step 203, the use behavior data can represent the use preference of the user for the candidate. In this way, the embodiment of the present application can obtain the user portrait feature reflecting the user preference according to the label of the candidate preferred by the user, and improve the accuracy of the user portrait feature.

[0097] The user portrait feature is also called user role, which can be used as an effective tool for sketching target users, contacting user appeals and designing directions. The user portrait feature can include gender, identity and other features, or the user portrait feature can include language style features, which can represent the language style features adopted by the user in the input process in the computer.

[0098] In the conventional technology, the user portrait feature is usually determined according to the personal information filled by the user. However, the personal information usually involves the privacy of the user, so the user is usually reluctant to fill in the personal information, and may fill in false personal information, resulting in errors in the user portrait feature.

[0099] Alternatively, in the conventional technology, the user portrait feature can be determined according to the purchase behavior of the user on the Internet. However, the goods involved in the purchase behavior are not necessarily goods purchased by the user himself. Therefore, the user portrait feature determined according to the purchase behavior also has errors.

[0100] The use behavior data of the user for the candidate is real data naturally accumulated by the user himself in the input process, so the accuracy of the user portrait feature can be improved.

[0101] In an optional embodiment of the present application, the determination of the user portrait feature corresponding to the user specifically includes: determining the use frequency corresponding to the label according to the use behavior data and the label of the candidate; and determining the user portrait feature corresponding to the user according to the use frequency.

[0102] Alternatively, the use frequency corresponding to the label can be determined according to the use frequency of the user for the candidate and the label of the candidate. For example, the "female label" corresponds to candidate A1, candidate A2, candidate A3, and the use frequency of the user for candidate A1, candidate A2, candidate A3 is use frequency F1, use frequency F2, use frequency F3, respectively. The use frequency F1, use frequency F2, use frequency F3 can be fused to obtain the use frequency corresponding to the "female label".

[0103] Optionally, in the embodiment of the application, the user portrait feature of the user can be determined according to the target label that meets the preset frequency condition. The preset frequency condition can include: the frequency of use is higher than a frequency threshold; or, in the case of sorting the labels in descending order of frequency of use, the top N (N can be a natural number) labels.

[0104] Optionally, in the embodiment of the application, the user portrait feature can include the target label that meets the preset frequency condition. Alternatively, the user portrait feature can be determined according to the matching degree between the target label that meets the preset frequency condition and the preset portrait feature, for example, the preset portrait feature that matches the target label is taken as the user portrait feature.

[0105] In summary, the recommendation method of the embodiment of the application can be used to determine the user portrait feature according to the use behavior data of the user for the candidate items provided by the input method and the labels of the candidate items. The use behavior data can represent the use preference of the user for the candidate items. In this way, the embodiment of the application can determine the user portrait feature that reflects the user preference according to the labels of the candidate items preferred by the user, thereby improving the accuracy of the user portrait feature.

[0106] Method embodiment two

[0107] Reference Figure 3 Fig. 2 shows a step flowchart of a recommendation method embodiment two of the application, which can specifically include the following steps:

[0108] Step 301: determining matching information between candidate content and a user portrait feature; the user portrait feature can be determined according to use behavior data of the user for candidate items provided by an input method and labels of the candidate items;

[0109] Step 302: recommending the candidate content according to the matching information.

[0110] The embodiment of the application can be used to recommend any type of candidate content. Since the embodiment of the application determines the user portrait feature according to the use behavior data of the user for the candidate items provided by the input method and the labels of the candidate items, the accuracy of the user portrait feature can be improved. Therefore, the embodiment of the application can improve the accuracy of the recommendation by recommending the candidate content according to the user portrait feature with higher accuracy.

[0111] In the case of the candidate content being the candidate items provided by the input method, since the user portrait feature can reflect the input preference of the user, the candidate items recommended by the embodiment of the application can meet the personalized input intention of the user.

[0112] The type of the candidate content can include: a candidate item of an input type provided by the input method, a candidate content of a commodity type, or a candidate content corresponding to a news type, etc. It can be understood that the embodiments of the present application do not limit the specific type of the candidate content.

[0113] Optionally, the candidate item recommended by the input method can include: an association candidate. The language unit corresponding to the association candidate can include: a word, a phrase, a sentence, etc. The association candidate can be a candidate content corresponding to the context in the input environment.

[0114] The determination process of the association candidate is described below.

[0115] According to an embodiment, the association candidate can be obtained by querying the multi-element relationship data according to the context. The multi-element relationship data can include binary and more than binary relationship data. The binary relationship, also known as 2-gram, is used to represent the probability of the occurrence of two elements in succession. The more than binary relationship is used to represent the probability of the occurrence of more than two elements in succession. The elements can include: a word, a phrase, a letter, a symbol, or an expression, etc.

[0116] In an optional embodiment of the present application, the multi-element relationship data can be represented by an association model. The type of the association model can include but is not limited to: a language model, a neural network model, etc. The above data model can provide P (any element | context, …), i.e. the probability of any element under the condition of a certain context. According to this probability, the hit element corresponding to the context can be determined, and thus the association candidate can be obtained. The corpus used by the association model can include: the corpus under the condition of the context.

[0117] Optionally, in the embodiments of the present application, a sentence association model can be trained using a sentence corpus, so that the sentence association model has a sentence-level association function; and the server determines the corresponding sentence association candidate for the context with high integrity according to the sentence association model, which can improve the relevance between the sentence association candidate and the input content, and thus can improve the input efficiency and enhance the user experience.

[0118] In an optional embodiment of the present application, the determination process of the association candidate can specifically include: determining a target language style corresponding to the context; and determining the sentence association candidate corresponding to the context by using the sentence association model corresponding to the target language style.

[0119] The embodiments of the present application can set different sentence association models for different target language styles. In this way, the sentence association candidate corresponding to the context can be determined by using the sentence association model corresponding to the target language style.

[0120] In another optional embodiment of the present application, the sentence association model determines the target language style corresponding to the context, and determines the sentence association candidate corresponding to the context according to the target language style. The embodiment of the present application can not distinguish the sentence association model according to the target language style, but the sentence association model first determines the target language style corresponding to the context in the processing of the context, and then determines the sentence association candidate corresponding to the context according to the target language style.

[0121] In an optional embodiment of the present application, the sentence association model can include a mapping relationship between the second vector and the sentence association candidate.

[0122] The determination manner of the sentence association candidate can include: determining a first vector corresponding to the context; searching in the mapping relationship between the second vector and the sentence association candidate according to the first vector to obtain the sentence association candidate corresponding to the first vector; the mapping relationship can be obtained according to a sentence corpus, the sentence corpus can include a context sample and a sentence association candidate, and the second vector corresponds to the context.

[0123] The embodiment of the present application obtains the sentence association candidate corresponding to the context based on vector searching; the first vector and the second vector can be matched based on the distance between the first vector and the second vector in the space, and the sentence association candidate corresponding to the second vector matched with the first vector is taken as the sentence association candidate corresponding to the context. In this way, the first vector and the second vector do not need to be strictly consistent in the text, so the coverage of the sentence association candidate can be increased.

[0124] For example, the sentence corpus includes the context sample A "Can you talk to you?" and the sentence association candidate "I want to hear your voice", and in the case that the user context a is "Can you talk to you?", the embodiment of the present application can determine that the first vector and the second vector are matched based on the spatial distance between the first vector and the second vector, and then provide the corresponding sentence association candidate "I want to hear your voice".

[0125] According to an embodiment, the first vector can be matched with all the second vectors corresponding to all the sentence corpora. Specifically, the distance between the first vector and all the second vectors can be calculated, and the second vector with a distance less than a first distance threshold is taken as the second vector matched with the first vector.

[0126] The distance measurement method between vectors can include Euclidean distance, cosine of included angle, Hamming distance, or Jaccard similarity coefficient, etc. It can be understood that the embodiment of the present application does not limit the specific distance measurement method between vectors.

[0127] According to another embodiment, the searching in the mapping relationship between the second vector and the sentence association candidate specifically comprises: determining a target index corresponding to the first vector according to the indexes of the first vector and the second vector; and determining the sentence association candidate corresponding to the first vector according to the second vector corresponding to the first vector and the target index.

[0128] The embodiment of the present application can pre-establish the indexes of the second vectors; thus, in the vector searching process, the target index corresponding to the first vector can be determined first, and then the second vector corresponding to the first vector and the target index is matched. Since the matching operation of the second vector corresponding to the non-target index can be saved, the vector searching efficiency can be improved.

[0129] An example of establishing the indexes of the second vectors is provided herein. The example specifically comprises: clustering the second vectors to obtain a plurality of vector categories; and establishing the indexes of the second vectors in the vector categories according to the information of the vector categories.

[0130] The information of the vector categories can comprise the center vectors of the vector categories. Accordingly, the determination of the target index corresponding to the first vector specifically comprises: determining the target vector category corresponding to the first vector according to the distance between the first vector and the center vector corresponding to the vector category. Alternatively, the vector category with a distance less than a second distance threshold can be taken as the target vector category corresponding to the first vector, and the target vector category corresponds to the target index.

[0131] The user portrait feature of the embodiment of the present application can comprise the features of gender and identity, or the user portrait feature can comprise the language style feature, which can represent the language style feature of the language used by the user in the process of inputting in the computer.

[0132] The method of the embodiment of the present application can be applied to a communication environment, and the user portrait feature can comprise at least one of the following features: the first user portrait feature of the local user, the second user portrait feature of at least one opposite user, and the user relationship feature of the multi-user. In the communication environment, the candidate content is usually the content sent to the opposite user, and thus the embodiment of the present application expands the second user portrait feature of the opposite user and the user relationship feature of the multi-user in the user portrait feature, which can improve the precision of the candidate content.

[0133] For example, in the case that the second user portrait feature is “female”, the candidate content conforming to “female” can be adopted; or in the case that the user relationship is “male”, the candidate content conforming to “male” can be adopted.

[0134] For example, in the case that the user relationship feature is "lover", the candidate content in "sweet style" can be adopted; or in the case that the user relationship is "colleague", the candidate content in "serious style" can be adopted.

[0135] In step 301, the matching information between the candidate content and the user portrait feature is determined, which can include determining the matching information between the label corresponding to the candidate content and the user portrait feature. For example, the matching degree between the label corresponding to the candidate content and the user portrait feature in the vector aspect can be determined by using the vector matching method. It can be understood that the specific process of determining the matching information between the candidate content and the user portrait feature is not limited in the embodiment of the application.

[0136] In step 302, the candidate content is recommended according to the matching information, and the candidate content that matches the user portrait feature more can be recommended to the user to improve the accuracy of the recommendation.

[0137] According to an embodiment, the candidate content is recommended, which specifically includes: sorting the multiple candidate contents according to the matching information; and displaying the multiple candidate contents according to the sorting result. For example, the multiple candidate contents can be sorted according to the order from high to low of the matching information, and the multiple candidate contents can be displayed according to the sorting result, so that the display position of the candidate content that matches the user portrait feature more is advanced.

[0138] According to another embodiment, the candidate content is recommended, which specifically includes: determining the target candidate content from the multiple candidate contents according to the matching information; and recommending the target candidate content.

[0139] In the embodiment of the application, the target candidate content can be obtained according to the candidate content that meets the preset matching condition. The preset matching condition can include: the matching information exceeding the matching threshold; or in the case that the multiple candidate contents are sorted according to the order from high to low of the matching information, the first N (N can be a natural number) candidate contents in the front.

[0140] In an application example of the application, in the recommendation scenario of the input method, the candidate items corresponding to the input content "I want to buy" specifically include "clothes", "this", "things", "lipstick", "game console", etc. In the traditional technology, the same candidate items are recommended for different users.

[0141] The embodiment of the present application can obtain the user portrait feature according to the user behavior data of the candidate item and the label of the candidate item. For example, if user A uses the candidate item with the label of "female" for many times, it can be considered that the user portrait feature of user A includes "female", and therefore, in the case that the on-screen content of user A is "I want to buy", the candidate item that meets "female" can be recommended, such as "clothes", "lipstick", etc. Further, if user A uses the candidate item with the label of "makeup" for many times, it can be considered that the user portrait feature of user A also includes "makeup", and therefore, the candidate item that meets "female" and "makeup" can be recommended, such as "lipstick", etc.

[0142] For another example, if user B uses the candidate item with the label of "male" and "electronic product" for many times, it can be considered that the user portrait feature of user B includes "male" and "electronic product", and therefore, in the case that the on-screen content of user B is "I want to buy", the candidate item that meets "male" and "electronic product" can be recommended, such as "game console", etc.

[0143] In another application example of the present application, if user C clicks the candidate item with the label of "cute style" such as "I want to buy Bujiasan" for many times, the embodiment of the present application can consider that the user portrait feature of user C includes "cute style". Therefore, in the input process of user C, the candidate item with "cute style" can be recommended to user C, or the display position of the candidate item with "cute style" can be advanced.

[0144] It should be noted that in the case of determining the user portrait feature, the recommendation method of the embodiment of the present application can be used for recommendation.

[0145] In the case of not determining the user portrait feature, the recommendation method of the traditional technology can be used for recommendation.

[0146] Alternatively, a random method can be used to determine the recommended label from the labels corresponding to the candidate content, and the candidate content corresponding to the recommended label is recommended, so as to improve the display probability of the candidate content of different labels based on multiple recommendations, and thus the accumulation of the user behavior data can be realized.

[0147] Alternatively, the candidate content corresponding to multiple labels can be recommended, for example, the candidate content corresponding to multiple labels can be displayed in one recommendation process, so as to improve the display probability of the candidate content of multiple labels based on one recommendation, and thus the accumulation of the user behavior data can be realized.

[0148] In summary, the recommendation method of the embodiment of the present application can improve the accuracy of the user portrait feature by obtaining the user portrait feature according to the use behavior data of the user for the candidate item provided by the input method and the label of the candidate item. Therefore, the embodiment of the present application can improve the accuracy of the recommendation by recommending the candidate content according to the user portrait feature with higher accuracy.

[0149] It should be noted that, for the method embodiment, in order to simply describe, it is expressed as a series of motion action combinations, but those skilled in the art should know that the embodiment of the present application is not limited by the described action sequence, because according to the embodiment of the present application, certain steps can be performed in other order or simultaneously. Secondly, those skilled in the art should know that the embodiments described in the specification all belong to preferred embodiments, and the motion actions involved are not necessarily required by the embodiment of the present application.

[0150] Device embodiment

[0151] Referring to Figure 4 , a structural block diagram of a recommendation device embodiment of the present application is shown, which can specifically include:

[0152] The determination module 401 is configured to determine the matching information between the candidate content and the user portrait feature; the user portrait feature can be obtained according to the use behavior data of the user for the candidate item provided by the input method and the label of the candidate item; and

[0153] The recommendation module 402 is configured to recommend the candidate content according to the matching information.

[0154] Optionally, the device is applied to a communication environment, and the user portrait feature can include at least one of the following features:

[0155] The first user portrait feature of the local user, the second user portrait feature of at least one opposite user, and the user relationship feature of the multi-user.

[0156] Optionally, the determination module 401 is specifically configured to determine the matching information between the label corresponding to the candidate content and the user portrait feature.

[0157] Optionally, the candidate content can be the candidate content corresponding to the context in the input environment.

[0158] Optionally, the recommendation module 402 can include:

[0159] The sorting module is configured to sort the plurality of candidate contents according to the matching information;

[0160] The display module is configured to display the plurality of candidate contents according to the obtained sorting result.

[0161] Optionally, the recommendation module 402 can comprise:

[0162] a target candidate content determination module configured to determine a target candidate content from the plurality of candidate contents according to the matching information;

[0163] a target candidate content recommendation module configured to recommend the target candidate content.

[0164] With reference to Figure 5 , a structural block diagram of an embodiment of a recommendation device of the present application is shown, which can specifically comprise:

[0165] a label adding module 501 configured to add a label to a candidate item provided by an input method;

[0166] a data determination module 502 configured to determine usage behavior data of a user with respect to the candidate item; and

[0167] a portrait determination module 503 configured to determine a user portrait feature of the user according to the usage behavior data and the label of the candidate item.

[0168] Optionally, the candidate item is applied to a communication environment, and the label can comprise at least one of the following features:

[0169] a first user label of a local user, at least one second user label of a remote user, and a user relationship label of a multi-user.

[0170] Optionally, the label adding module 501 can comprise:

[0171] a first label adding module configured to determine the label corresponding to the candidate item according to semantic information corresponding to the candidate item; and / or

[0172] a second label adding module configured to determine the label corresponding to the candidate item according to a classification model corresponding to the label; and / or

[0173] a third label adding module configured to determine the label corresponding to the candidate item according to a mapping relationship between the candidate item and the label.

[0174] Optionally, the device can further comprise:

[0175] a candidate item construction module configured to construct a corresponding candidate item with respect to a preset label, and establish a mapping relationship between the candidate item and the label.

[0176] Optionally, the candidate item construction module can comprise:

[0177] a first candidate item construction module configured to construct a corresponding candidate item according to a preset label and a preset context; and / or

[0178] The second candidate construction module is configured to construct a second candidate corresponding to a second language style label according to the first language style label and the first candidate corresponding to the first language style label, and the second language style label.

[0179] Optionally, the portrait determination module 503 can include:

[0180] The frequency determination module is configured to determine a use frequency corresponding to the label according to the use behavior data and the label of the candidate.

[0181] The portrait feature determination module is configured to determine a user portrait feature corresponding to the user according to the use frequency.

[0182] For the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the related parts refer to the part of the method embodiment.

[0183] Each embodiment in the specification is described in a progressive manner, and each embodiment focuses on the difference from other embodiments. The same and similar parts between the embodiments can be referred to each other.

[0184] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment related to the method, and will not be described in detail here.

[0185] The embodiment of the application provides a device for recommendation, including a memory, and one or more programs, wherein the one or more programs are stored in the memory, and are configured to be executed by one or more processors, and the one or more programs include instructions for determining matching information between candidate content and user portrait features; the user portrait features are obtained according to user use behavior data for input method provided candidate and labels of the candidate; and the candidate content is recommended according to the matching information.

[0186] Figure 6 Fig. 8 is a block diagram of a device 800 for recommendation according to an example embodiment. For example, the device 800 can be a mobile phone, a computer, a digital broadcast terminal, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, etc.

[0187] Referring to Figure 6The device 800 can include one or more of the following components: a processing component 802, a memory 804, a power supply component 806, a multimedia component 808, an audio component 810, an input / output (I / O) interface 812, a sensor component 814, and a communication component 816.

[0188] The processing component 802 generally controls the overall operation of the device 800 such as the operation of the display, the telephone call, the data communication, the camera operation and the recording operation. The processing component 802 can include one or more processors 820 to execute instructions to complete all or a part of steps of the methods described above. Furthermore, the processing component 802 can include one or more modules to facilitate interaction between the processing component 802 and other components. For example, the processing component 802 can include a multimedia module to facilitate the interaction between the multimedia component 808 and the processing component 802.

[0189] The memory 804 is configured to store various types of data to support the operations of the device 800. Examples of these data include instructions to execute any applications or methods on the device 800, contact data, phonebook data, messages, pictures, videos and so on. The memory 804 can be implemented by any type of volatile or non-volatile storage devices or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0190] The power supply component 806 supplies the power for the various components of the device 800. The power supply component 806 can include a power supply management system, one or more power supplies, and other components associated with generating, managing and distributing power for the device 800.

[0191] The multimedia component 808 includes a screen to provide an output interface between the device 800 and a user. In some embodiments, the screen can include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes the touch panel, the screen can be implemented as a touch screen to receive an input signal from a user. The touch panel includes one or more touch sensors to sense a touch, a slide and a gesture on the touch panel. The touch sensors can not only sense a boundary of a touching or sliding action, but also detect duration and pressure related to the touching or sliding action. In some embodiments, the multimedia component 808 includes a front camera and / or a rear camera. The front and / or rear camera can receive external multimedia data when the device 800 is in an operation mode, such as a shooting mode or a video mode. Each of the front and rear camera can be a fixed optical lens system or have a focal length and optical zoom capability.

[0192] The audio component 810 is configured to output and / or input audio signals. For example, the audio component 810 includes a microphone (MIC) that is configured to receive an external audio signal when the device 800 is in an operation mode, such as a call mode, a recording mode, and a voice recognition mode. The received audio signal can be further stored in the memory 804 or transmitted via the communication component 816. In some embodiments, the audio component 810 also includes a speaker for outputting audio signals.

[0193] The I / O interface 812 provides an interface between the processing component 802 and peripheral interface modules, which can include a keypad, a click wheel, buttons, and so on. The buttons can include, but are not limited to, a home button, a volume button, a start button, and a lock button.

[0194] The sensor component 814 includes one or more sensors for providing status assessments of various aspects of the device 800. For example, the sensor component 814 can detect an open / closed position of the device 800, relative positioning of components, such as a display and a keypad of the device 800, a change of position of the device 800 or a component of the device 800, presence or absence of user contact with the device 800, changes in orientation or acceleration / deceleration / velocity of the device 800, and temperature changes of the device 800, among a plethora of other examples. The sensor component 814 can include proximity sensor configured to detect presence of an object in a proximity without any physical touch. The sensor component 814 can also include a light sensor (e.g., a CMOS or CCD image sensor) configured to work in an imaging application. In some embodiments, the sensor component 814 can also include an acceleration sensor, a gyroscope sensor, a magnetic sensor, a pressure sensor, or a temperature sensor.

[0195] The communication component 816 is configured to facilitate wired or wireless communication between the device 800 and other devices. The device 800 can access a wireless network based on a corresponding communication standard, such as WiFi, 2G, or 3G, or a combination thereof. In an example embodiment, the communication component 816 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In an example embodiment, the communication component 816 further includes a Near Field Communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on Radio Frequency Identification (RFID) technology, infrared data association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology and other technology.

[0196] In exemplary embodiments, the apparatus 800 can be implemented using one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, micro-controllers, microprocessors or other electronic devices, to perform the above methods.

[0197] In exemplary embodiments, a non-transitory computer readable storage medium including instructions, such as the memory 804 including instructions, is also provided, which can be executed by the processor 820 of the apparatus 800 to complete the above methods. For example, the non-transitory computer readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disc, and an optical data storage device, etc.

[0198] Figure 7 is a schematic diagram of a server in some embodiments of the present application. The server 1900 can vary greatly due to different configurations or performances, and can include one or more central processing units (CPUs) 1922 (e.g., one or more processors) and a memory 1932, one or more storage media 1930 (e.g., one or more mass storage devices) storing application programs 1942 or data 1944. Among them, the memory 1932 and the storage medium 1930 can be temporary storage or persistent storage. The programs stored in the storage medium 1930 can include one or more modules (not shown in the figure), each of which can include a series of instruction operations in the server. Further, the central processing unit 1922 can be configured to communicate with the storage medium 1930 and execute a series of instruction operations in the storage medium 1930 on the server 1900.

[0199] The server 1900 can also include one or more power supplies 1926, one or more wired or wireless network interfaces 1950, one or more input / output interfaces 1958, one or more keyboards 1956, and / or one or more operating systems 1941, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.

[0200] A non-transitory computer readable storage medium, when the instructions in the storage medium are executed by the processor of the apparatus (server or terminal), enable the apparatus to perform Figure 2 or Figure 3 the recommended method shown.

[0201] A non-transitory computer readable storage medium, when instructions in the storage medium are executed by a processor of an apparatus (server or terminal), enables the apparatus to perform a recommendation method, the method comprising: determining matching information between candidate content and user portrait features; the user portrait features being obtained according to user usage behavior data for candidate items provided by an input method and labels of the candidate items; and recommending the candidate content according to the matching information.

[0202] Embodiments of the present application disclose A1, a recommendation method, the method comprising:

[0203] determining matching information between candidate content and user portrait features; the user portrait features being obtained according to user usage behavior data for candidate items provided by an input method and labels of the candidate items;

[0204] recommending the candidate content according to the matching information.

[0205] A2, the method according to A1, the method being applied to a communication environment, and the user portrait features comprising at least one of the following features:

[0206] first user portrait features of a local user, second user portrait features of at least one opposite user, and user relationship features of multi-user.

[0207] A3, the method according to A1, the determining of the matching information between the candidate content and the user portrait features comprising:

[0208] determining matching information between labels corresponding to the candidate content and the user portrait features.

[0209] A4, the method according to any one of A1 to A3, the candidate content being candidate content corresponding to a context in an input environment.

[0210] A5, the method according to any one of A1 to A3, the recommending of the candidate content comprising:

[0211] ranking a plurality of candidate contents according to the matching information;

[0212] displaying the plurality of candidate contents according to a ranking result obtained.

[0213] A6, the method according to any one of A1 to A3, the recommending of the candidate content comprising:

[0214] determining target candidate content from the plurality of candidate contents according to the matching information;

[0215] recommending the target candidate content.

[0216] The embodiment of the present application discloses B7, a recommendation method, which comprises:

[0217] adding a label to the candidate provided for the input method;

[0218] determining the use behavior data of the user for the candidate;

[0219] determining the user portrait feature corresponding to the user according to the use behavior data and the label of the candidate.

[0220] B8, according to B7, the method, the candidate is applied to the communication environment, and the label comprises at least one of the following features:

[0221] The first user label of the local user, the second user label of at least one opposite user and the user relationship label of the multi-user.

[0222] B9, according to B7, the method, the label is added to the candidate provided for the input method, comprising:

[0223] determining the label corresponding to the candidate according to the semantic information corresponding to the candidate; and / or

[0224] determining the label corresponding to the candidate according to the classification model corresponding to the label; and / or

[0225] determining the label corresponding to the candidate according to the mapping relationship between the candidate and the label.

[0226] B10, according to B9, the method further comprises:

[0227] constructing the corresponding candidate for the preset label, and establishing the mapping relationship between the candidate and the label.

[0228] B11, according to B10, the method, the corresponding candidate is constructed, comprising:

[0229] constructing the corresponding candidate according to the preset label and the preset context; and / or

[0230] constructing the second candidate corresponding to the second language style label according to the first language style label and the first candidate corresponding thereto, and the second language style label,

[0231] B12, according to any one of B7 to B11, the method, the user portrait feature corresponding to the user is determined, comprising:

[0232] determining the use frequency corresponding to the label according to the use behavior data and the label of the candidate;

[0233] According to the use frequency, a user portrait feature corresponding to the user is determined.

[0234] Embodiments of the present application disclose C13, a recommendation device, comprising:

[0235] A determination module is configured to determine matching information between candidate content and a user portrait feature, wherein the user portrait feature is obtained according to user use behavior data for input method provided candidate items and labels of the candidate items; and

[0236] A recommendation module is configured to recommend the candidate content according to the matching information.

[0237] C14, the device of C13, the device is applied to a communication environment, and the user portrait feature comprises at least one of the following features:

[0238] A first user portrait feature of a local user, a second user portrait feature of at least one opposite user and a user relationship feature of a multi-user.

[0239] C15, the device of C13, and the determination module is specifically configured to determine matching information between a label corresponding to candidate content and a user portrait feature.

[0240] C16, the device of any one of C13 to C15, and the candidate content is candidate content corresponding to a context in an input environment.

[0241] C17, the device of any one of C13 to C15, and the recommendation module comprises:

[0242] A sorting module is configured to sort a plurality of candidate contents according to the matching information.

[0243] A display module is configured to display the plurality of candidate contents according to a sorting result obtained.

[0244] C18, the device of any one of C13 to C15, and the recommendation module comprises:

[0245] A target candidate content determination module is configured to determine target candidate content from the plurality of candidate contents according to the matching information.

[0246] A target candidate content recommendation module is configured to recommend the target candidate content.

[0247] Embodiments of the present application disclose D19, a recommendation device, comprising:

[0248] A label adding module is configured to add labels to input method provided candidate items.

[0249] a data determining module configured to determine user usage behavior data of the user for the candidate item; and

[0250] an image determining module configured to determine a user image feature of the user according to the usage behavior data and a label of the candidate item.

[0251] D20. The apparatus of D19, wherein the candidate item is applied to a communication environment, and the label comprises at least one of the following features:

[0252] a first user label of a local user, a second user label of at least one remote user, and a user relationship label of a multi-user.

[0253] D21. The apparatus of D19, wherein the label adding module comprises:

[0254] a first label adding module configured to determine the label of the candidate item according to semantic information of the candidate item; and / or

[0255] a second label adding module configured to determine the label of the candidate item according to a classification model corresponding to the label; and / or

[0256] a third label adding module configured to determine the label of the candidate item according to a mapping relationship between the candidate item and the label.

[0257] D22. The apparatus of D21, further comprising:

[0258] a candidate item constructing module configured to construct a corresponding candidate item for a preset label and establish a mapping relationship between the candidate item and the label.

[0259] D23. The apparatus of D22, wherein the candidate item constructing module comprises:

[0260] a first candidate item constructing module configured to construct a corresponding candidate item according to a preset label and a preset context; and / or

[0261] a second candidate item constructing module configured to construct a second candidate item corresponding to a second language style label according to the first language style label and a first candidate item corresponding to the first language style label, and the second language style label,

[0262] D24. The apparatus of any one of D19 to D23, wherein the image determining module comprises:

[0263] a frequency determining module configured to determine a usage frequency of the label according to the usage behavior data and the label of the candidate item;

[0264] an image feature determining module configured to determine a user image feature of the user according to the usage frequency.

[0265] E25, An apparatus for recommendation, comprising a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by one or more processors, the one or more programs comprising instructions for:

[0266] determining matching information between the candidate content and user portrait features, wherein the user portrait features are obtained according to user usage behavior data for candidate items provided by an input method and labels of the candidate items;

[0267] recommending the candidate content according to the matching information.

[0268] E26, The apparatus of E25, applied to a communication environment, wherein the user portrait features comprise at least one of the following features:

[0269] first user portrait features of a local user, second user portrait features of at least one remote user, and user relationship features of multi-user.

[0270] E27, The apparatus of E25, wherein the determining of the matching information between the candidate content and the user portrait features comprises:

[0271] determining matching information between labels corresponding to the candidate content and the user portrait features.

[0272] E28, The apparatus of any one of E25 to E27, wherein the candidate content is candidate content corresponding to a context in an input environment.

[0273] E29, The apparatus of any one of E25 to E27, wherein the recommending of the candidate content comprises:

[0274] sorting a plurality of candidate contents according to the matching information;

[0275] displaying the plurality of candidate contents according to a sorting result obtained.

[0276] E30, The apparatus of any one of E25 to E27, wherein the recommending of the candidate content comprises:

[0277] determining a target candidate content from the plurality of candidate contents according to the matching information;

[0278] recommending the target candidate content.

[0279] Embodiments of the present application disclose a device for recommendation, comprising a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by one or more processors, and the one or more programs contain instructions for performing the following operations:

[0280] adding a label to a candidate provided for an input method;

[0281] determining usage behavior data of a user for the candidate;

[0282] determining a user portrait feature corresponding to the user according to the usage behavior data and the label of the candidate.

[0283] F32, the device according to F31, the candidate is applied to a communication environment, and the label comprises at least one of the following features:

[0284] a first user label of a local user, a second user label of at least one opposite user, and a user relationship label of a multi-user.

[0285] F33, the device according to F31, the adding of the label to the candidate provided for the input method comprises:

[0286] determining the label corresponding to the candidate according to semantic information corresponding to the candidate; and / or

[0287] determining the label corresponding to the candidate according to a classification model corresponding to the label; and / or

[0288] determining the label corresponding to the candidate according to a mapping relationship between the candidate and the label.

[0289] F34, the device according to F33, the device is further configured to execute the one or more programs by the one or more processors, and the one or more programs contain instructions for performing the following operations:

[0290] constructing a corresponding candidate for a preset label, and establishing a mapping relationship between the candidate and the label.

[0291] F35, the device according to F34, the constructing of the corresponding candidate comprises:

[0292] constructing the corresponding candidate according to the preset label and a preset context; and / or

[0293] constructing a second candidate corresponding to a second language style label according to a first language style label and a first candidate corresponding to the first language style label, and the second language style label,

[0294] F36. The apparatus of any one of F31-F35, wherein the determining the user portrait feature corresponding to the user comprises:

[0295] determining, according to the usage behavior data and the label of the candidate item, a usage frequency corresponding to the label;

[0296] determining, according to the usage frequency, a user portrait feature corresponding to the user.

[0297] Embodiments of the present application disclose G37, a machine readable medium having instructions stored thereon that, when executed by one or more processors, cause an apparatus to perform the recommendation method of any one of A1-A6.

[0298] Embodiments of the present application disclose H38, a machine readable medium having instructions stored thereon that, when executed by one or more processors, cause an apparatus to perform the recommendation method of any one of B7-B12.

[0299] Other embodiments of the application will be apparent to those skilled in the art from consideration of the specification and practice of the application disclosed herein. It is intended that the specification and examples be considered as exemplary only, with the true scope and spirit of the application being indicated by the following claims.

[0300] It should be understood that the application is not limited to the precise construction and compositions that have been described above and shown in the accompanying drawings and that variations and modifications exist within the scope and spirit of the application. The scope of the application is limited only by the claims that follow.

[0301] The above description is only preferred embodiments of the present application, and is not used to limit the present application, and any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application should be included in the protection scope of the present application.

[0302] The above provides a recommendation method, a recommendation device and a device for recommendation, and the principles and implementation modes of the present application are described by applying specific examples; the above description of the embodiments is only used to help understand the method and core idea of the present application; meanwhile, for those skilled in the art, according to the idea of the present application, the specific implementation modes and application ranges will have changes, and the above description should not be understood as limiting the present application.

Claims

1. A recommendation method characterized by comprising: The method is applied to a communication environment, and the method comprises: determining matching information between candidate content and user portrait features; the user portrait features are obtained according to user usage behavior data for input method provided candidate items and labels of the candidate items, and the adding mode of the labels comprises: determining labels corresponding to the candidate items according to a mapping relationship between the candidate items and the labels, wherein the adding mode comprises: for a preset label, constructing corresponding candidate items and establishing a mapping relationship between the candidate items and the labels; the constructing of the corresponding candidate items for the preset label comprises: constructing the corresponding candidate items according to the preset label and a preset context; and / or constructing second candidate items corresponding to a second language style label according to a first language style label and first candidate items corresponding to the first language style label and the second language style label; the usage behavior data comprises on-screen data of the candidate items; the user portrait features comprise at least one of the following features: first user portrait features of a local user, second user portrait features of at least one opposite user and user relationship features of multi-terminal users; determining matching information between candidate content and user portrait features; the user portrait features are obtained according to user usage behavior data for input method provided candidate items and labels of the candidate items, and the adding mode of the labels comprises: determining labels corresponding to the candidate items according to a mapping relationship between the candidate items and the labels, wherein the adding mode comprises: for a preset label, constructing corresponding candidate items and establishing a mapping relationship between the candidate items and the labels; the constructing of the corresponding candidate items for the preset label comprises: constructing the corresponding candidate items according to the preset label and a preset context; and / or constructing second candidate items corresponding to a second language style label according to a first language style label and first candidate items corresponding to the first language style label and the second language style label; the usage behavior data comprises on-screen data of the candidate items; the user portrait features comprise at least one of the following features: first user portrait features of a local user, second user portrait features of at least one opposite user and user relationship features of multi-terminal users; 2. The method of claim 1, wherein, determining matching information between candidate content and user portrait features; the user portrait features are obtained according to user usage behavior data for input method provided candidate items and labels of the candidate items, and the adding mode of the labels comprises: determining labels corresponding to the candidate items according to a mapping relationship between the candidate items and the labels, wherein the adding mode comprises: for a preset label, constructing corresponding candidate items and establishing a mapping relationship between the candidate items and the labels; the constructing of the corresponding candidate items for the preset label comprises: constructing the corresponding candidate items according to the preset label and a preset context; and / or constructing second candidate items corresponding to a second language style label according to a first language style label and first candidate items corresponding to the first language style label and the second language style label; the usage behavior data comprises on-screen data of the candidate items; the user portrait features comprise at least one of the following features: first user portrait features of a local user, second user portrait features of at least one opposite user and user relationship features of multi-terminal users; determining matching information between candidate content and user portrait features; the user portrait features are obtained according to user usage behavior data for input method provided candidate items and labels of the candidate items, and the adding mode of the labels comprises: determining labels corresponding to the candidate items according to a mapping relationship between the candidate items and the labels, wherein the adding mode comprises: for a preset label, constructing corresponding candidate items and establishing a mapping relationship between the candidate items and the labels; the constructing of the corresponding candidate items for the preset label comprises: constructing the corresponding candidate items according to the preset label and a preset context; and / or constructing second candidate items corresponding to a second language style label according to a first language style label and first candidate items corresponding to the first language style label and the second language style label; the usage behavior data comprises on-screen data of the candidate items; the user portrait features comprise at least one of the following features: first user portrait features of a local user, second user portrait features of at least one opposite user and user relationship features of multi-terminal users; 3. A recommendation method characterized by, determining matching information between candidate content and user portrait features; the user portrait features are obtained according to user usage behavior data for input method provided candidate items and labels of the candidate items, and the adding mode of the labels comprises: determining labels corresponding to the candidate items according to a mapping relationship between the candidate items and the labels, wherein the adding mode comprises: for a preset label, constructing corresponding candidate items and establishing a mapping relationship between the candidate items and the labels; the constructing of the corresponding candidate items for the preset label comprises: constructing the corresponding candidate items according to the preset label and a preset context; and / or constructing second candidate items corresponding to a second language style label according to a first language style label and first candidate items corresponding to the first language style label and the second language style label; the usage behavior data comprises on-screen data of the candidate items; the user portrait features comprise at least one of the following features: first user portrait features of a local user, second user portrait features of at least one opposite user and user relationship features of multi-terminal users; determining matching information between candidate content and user portrait features; the user portrait features are obtained according to user usage behavior data for input method provided candidate items and labels of the candidate items, and the adding mode of the labels comprises: determining labels corresponding to the candidate items according to a mapping relationship between the candidate items and the labels, wherein the adding mode comprises: for a preset label, constructing corresponding candidate items and establishing a mapping relationship between the candidate items and the labels; the constructing of the corresponding candidate items for the preset label comprises: constructing the corresponding candidate items according to the preset label and a preset context; and / or constructing second candidate items corresponding to a second language style label according to a first language style label and first candidate items corresponding to the first language style label and the second language style label; the usage behavior data comprises on-screen data of the candidate items; the user portrait features comprise at least one of the following features: first user portrait features of a local user, second user portrait features of at least one opposite user and user relationship features of multi-terminal users; ​ 4. A recommendation device characterized by comprising: ​ ​ The user portrait feature is obtained according to user usage behavior data for a candidate item provided by an input method and a label of the candidate item, and the label adding manner comprises: determining a label corresponding to the candidate item according to a mapping relationship between the candidate item and the label, wherein the determining comprises: for a preset label, constructing a corresponding candidate item and establishing a mapping relationship between the candidate item and the label; the constructing of the corresponding candidate item for the preset label comprises: constructing a corresponding candidate item according to the preset label and a preset context; and / or constructing a second candidate item corresponding to a second language style label according to a first language style label and a first candidate item corresponding to the first language style label and the second language style label; the usage behavior data comprises: on-screen data of the candidate item; the user portrait feature comprises at least one of the following features: a first user portrait feature of a local user, a second user portrait feature of at least one opposite user and a user relationship feature of multi-terminal users; and The recommendation module is configured to recommend the candidate content according to the matching information.

5. A recommendation device characterized by comprising: The apparatus comprises: The label adding module is configured to add a label to a candidate item provided by an input method, wherein the adding comprises: determining a label corresponding to the candidate item according to a mapping relationship between the candidate item and the label, wherein the determining comprises: for a preset label, constructing a corresponding candidate item and establishing a mapping relationship between the candidate item and the label; the constructing of the corresponding candidate item for the preset label comprises: constructing a corresponding candidate item according to the preset label and a preset context; and / or constructing a second candidate item corresponding to a second language style label according to a first language style label and a first candidate item corresponding to the first language style label and the second language style label; The data determining module is configured to determine user usage behavior data for the candidate item; the usage behavior data comprises: on-screen data of the candidate item; and The portrait determining module is configured to determine a user portrait feature corresponding to the user according to the usage behavior data and the label of the candidate item; the user portrait feature comprises at least one of the following features: a first user portrait feature of a local user, a second user portrait feature of at least one opposite user and a user relationship feature of multi-terminal users.

6. An apparatus for recommendation, the apparatus comprising: The one or more programs stored in the memory are configured to be executed by the one or more processors, and the one or more programs contain instructions for performing the following operations: determine matching information between the candidate content and a user portrait feature; the user portrait feature is obtained according to user usage behavior data for a candidate item provided by an input method and a label of the candidate item, and the label is added in the following manner: a label corresponding to a candidate item is determined according to a mapping relationship between the candidate item and the label, which includes: for a preset label, constructing a corresponding candidate item and establishing a mapping relationship between the candidate item and the label; the constructing of the corresponding candidate item for the preset label includes: constructing the corresponding candidate item according to the preset label and a preset context; and / or constructing a second candidate item corresponding to a second language style label according to a first language style label and a first candidate item corresponding to the first language style label, and the second language style label; the usage behavior data includes: on-screen data of the candidate item; and the user portrait feature includes at least one of the following features: a first user portrait feature of a local user, a second user portrait feature of at least one opposite user, and a user relationship feature of multi-terminal users; recommend the candidate content according to the matching information.

7. An apparatus for recommendation, the apparatus comprising: include a memory and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by one or more processors, and the one or more programs contain instructions for performing the following operations: add a label to a candidate item provided by an input method, which includes: determining a label corresponding to a candidate item according to a mapping relationship between the candidate item and the label, which includes: for a preset label, constructing a corresponding candidate item and establishing a mapping relationship between the candidate item and the label; the constructing of the corresponding candidate item for the preset label includes: constructing the corresponding candidate item according to the preset label and a preset context; and / or constructing a second candidate item corresponding to a second language style label according to a first language style label and a first candidate item corresponding to the first language style label, and the second language style label; determine user usage behavior data for the candidate item; the usage behavior data includes: on-screen data of the candidate item; determine a user portrait feature corresponding to the user according to the usage behavior data and the label of the candidate item; the user portrait feature includes at least one of the following features: a first user portrait feature of a local user, a second user portrait feature of at least one opposite user, and a user relationship feature of multi-terminal users; find candidate content in a source database according to a context and / or an input string, the context including: on-screen content of a local user and / or communication content sent by an opposite user.

8. A machine-readable medium having instructions stored thereon that, when executed by one or more processors, cause an apparatus to perform the recommendation method of one or more of claims 1 to 2.

9. A machine-readable medium having instructions stored thereon that, when executed by one or more processors, cause an apparatus to perform the recommendation method of claim 3.

Images