CN105095401A - Method and apparatus for identifying gender - Google Patents

Abstract

The invention provides a method for identifying gender, which comprises: acquire voice datum of a plurality of terminals within a first preset time period and identifying the gender of a user corresponding to the plurality of terminals based on the voice datum; acquire travel datum of a plurality of terminals that have already identified the gender of the user within a second preset time period; use the travel datum of the plurality of terminals as training corpus to perform binary classification training to acquire a binary prediction model used for predicting the gender; and predict the gender of the user corresponding to the terminals based on the binary prediction model and the travel datum of the terminals. The invention also provides an apparatus for identifying the gender, which comprises a first gender identification unit, an acquisition unit, a prediction model establishing unit and a gender prediction unit. By adopting the invention, the prediction to the gender is fully analyzed; and the accuracy is high, thus further promoting different individualized products or applications for users of different genders based on the preferences of the different genders.

Description

A kind of sex knows method for distinguishing and device

Technical field

The present invention relates to computer processing technology field, particularly relate to a kind of sex and know method for distinguishing and device.

Background technology

At present; app application comprises the software of multiple social class; and due to the preference difference of the user of different sexes comparatively large, therefore various social class software can go to push the product of some personalizations and application etc. according to the sex of user usually, and the sex therefore identifying user is most important.

At present, identify that user's property method for distinguishing mainly comprises: by the data of the user individual aspects such as voice, user images and user preference data, identifiable design goes out the sex of user.But the individuation data of existing user only includes speech data or view data, but the speech data of same terminal or view data can be sent by multiple user, and therefore the sex of the user that this terminal is corresponding is difficult to be determined.

Summary of the invention

For user individual data deficiency in prior art terminal, the very doubt defect of the user's sex causing terminal corresponding, the invention provides a kind of sex and knows method for distinguishing and device.

First aspect, the invention provides a kind of sex and know method for distinguishing, the method comprises:

Obtain the speech data of multiple terminal in the first preset time period, and according to described speech data, the sex of user corresponding to described multiple terminal is identified;

Obtain in the second preset time period the trip data of the multiple terminals having identified user's sex;

The trip data of described multiple terminal is carried out two classification based trainings as corpus, obtains the two classification forecast models for gender prediction;

According to described two classification forecast models, and the trip data of terminal, the user sex corresponding to terminal is predicted.

Preferably, described according to described speech data, the sex of user corresponding to described multiple terminal is identified, comprising:

If the speech data of a terminal comprises a kind of speech data of sex, then using the sex of this sex as user corresponding to described terminal;

If the speech data of a terminal comprises the speech data of two kinds of sexes, then filter out the speech data that this terminal is corresponding.

Preferably, described trip data is go out line frequency, travel time section distribution, the POI distribution of trip purpose ground and frequency, user integral store redemption data week, give tip amount, give tip frequency, by certificate ratio and with one or more in certificate mean value.

Preferably, described according to described speech data, after the sex of user corresponding to described multiple terminal is identified, also comprise:

Obtain and identified user's friend relation corresponding to the terminal of user's sex;

According to described friend relation, obtain and described user good friend and the highest first user of pre-set priority each other;

Judge described first user and the described user having identified sex different in nature each other.

Preferably, the two classification forecast models obtained described in for gender prediction are:

$\underset{i}{\Σ}\[y_{i}LogP(y=1|x)+(1-y_i)LogP(y=0|x)\]$

Wherein, $P(y=1|x)=\frac{\exp \left(wx\right)}{1+\exp \left(wx\right)},P(y=0|x)=\frac{1}{1+\exp \left(wx\right)};$

Wherein, user's trip data that x represents, y=1 represents and is predicted as the male sex, and y=0 represents and is predicted as women, and w represents coefficient.

Second aspect, the invention provides a kind of device of sex identification, this device comprises:

First sex recognition unit, for obtaining the speech data of multiple terminal in the first preset time period, and according to described speech data, identifies the sex of user corresponding to described multiple terminal;

Acquiring unit, for obtaining in the second preset time period the trip data of the multiple terminals having identified user's sex;

Forecast model sets up unit, for the trip data of described multiple terminal is carried out two classification based trainings as corpus, obtains the two classification forecast models for gender prediction;

Gender prediction's unit, for according to described two classification forecast models, and the trip data of terminal, the user sex corresponding to terminal is predicted.

Preferably, described first sex recognition unit, for:

If the speech data of a terminal comprises a kind of speech data of sex, then using the sex of this sex as user corresponding to described terminal;

If the speech data of a terminal comprises the speech data of two kinds of sexes, then filter out the speech data that this terminal is corresponding.

Preferably, described trip data is go out line frequency, travel time section distribution, the POI distribution of trip purpose ground and frequency, user integral store redemption data week, give tip amount, give tip frequency, by certificate ratio and with one or more in certificate mean value.

Preferably, described device also comprises second sex recognition unit, for:

Obtain and identified user's friend relation corresponding to the terminal of user's sex;

According to described friend relation, obtain and described user good friend and the highest first user of pre-set priority each other;

Judge described first user and the described user having identified sex different in nature each other.

Preferably, the described two classification forecast models for sex test are:

$\underset{i}{\Σ}\[y_{i}LogP(y=1|x)+(1-y_i)LogP(y=0|x)\]$

Wherein, $P(y=1|x)=\frac{\exp \left(wx\right)}{1+\exp \left(wx\right)},P(y=0|x)=\frac{1}{1+\exp \left(wx\right)};$

Wherein, user's trip data that x represents, y=1 represents and is predicted as the male sex, and y=0 represents and is predicted as women, and w represents coefficient.。

As shown from the above technical solution, the invention provides a kind of sex and know method for distinguishing and device, identify according to the sex of speech data to user corresponding to part terminal, trip data training according to this part terminal obtains two disaggregated models, does not have the sex of user corresponding to the terminal of speech data to identify to other.Integrated voice of the present invention, trip data and friend relation determine the sex of user corresponding to terminal, analyze comprehensively, and accuracy are higher.Thus further according to the preference of different sexes, the user for different sexes pushes different personalized products or application.

Accompanying drawing explanation

In order to be illustrated more clearly in disclosure embodiment or technical scheme of the prior art, be briefly described to the accompanying drawing used required in embodiment or description of the prior art below, apparently, accompanying drawing in the following describes is only embodiments more of the present disclosure, for those of ordinary skill in the art, under the prerequisite not paying creative work, other accompanying drawing can also be obtained according to these figure.

Fig. 1 is the schematic flow sheet of a kind of sex knowledge method for distinguishing that the disclosure one embodiment provides;

Fig. 2 is the structural representation of the device of a kind of sex identification that another embodiment of the disclosure provides.

Embodiment

Below in conjunction with the accompanying drawing in disclosure embodiment, be clearly and completely described the technical scheme in disclosure embodiment, obviously, described embodiment is only disclosure part embodiment, instead of whole embodiments.Based on the embodiment in the disclosure, those of ordinary skill in the art are not making the every other embodiment obtained under creative work prerequisite, all belong to the scope of disclosure protection.

As shown in Figure 1, be the schematic flow sheet of a kind of sex knowledge method for distinguishing that the disclosure one embodiment provides, the method comprises the steps:

S1: the speech data obtaining multiple terminal in the first preset time period, and according to described speech data, identifies the sex of user corresponding to described multiple terminal.

Wherein, terminal refers to the terminal being provided with the application of social class, as mobile phone, PC etc.For example, speech data can be voice in the system of calling a taxi forms data.After speech data is extracted from database, according to the call format of the speech recognition program preset, format conversion is carried out to speech data, then adopt default speech recognition program to process speech data, obtain the identification of user corresponding to speech data.Then in the system of calling a taxi, described terminal for providing service side, as the driver in vehicles dial-a-cab, the equipment such as the mobile terminal for order used or personal computer (PersonalComputer is called for short PC).Such as smart mobile phone, personal digital assistant (PDA), panel computer, notebook computer, vehicle-mounted computer (carputer), handheld device, intelligent glasses, intelligent watch, wearable device, virtual display device or display enhancing equipment (as GoogleGlass, OculusRift, Hololens, GearVR) etc.

S2: the trip data obtaining in the second preset time period the multiple terminals having identified user's sex.

Specifically, identify the sex of user corresponding to multiple terminal according to step S1 after, obtain the trip data of above-mentioned multiple terminal further, for training two disaggregated models.

For example, in the system of calling a taxi, described trip data can be to go out line frequency, travel time section distribution, the POI distribution of trip purpose ground and frequency, user integral store redemption data week, give tip amount, give tip frequency, by certificate ratio and with one or more in certificate mean value.

S3: the trip data of described multiple terminal is carried out two classification based trainings as corpus, obtains the two classification forecast models for gender prediction.

Wherein, two classification in this step refer to logistic regression, and concrete training method is:

Using gender prediction's result of speech data as training input data; Everyone trip data of sex will be determined by speech data as a line training sample, utilize logistic regression training program to carry out conventional model training.

Specifically, using gender prediction's result as training input data y, trip data is trained the logistic regression objective function shown in formula (1) as training sample x, tries to achieve coefficient w, thus obtain two classification forecast models:

$\underset{i}{\Σ}\[y_{i}LogP(y=1|x)+(1-y_i)LogP(y=0|x)\]$

Wherein, $P(y=1|x)=\frac{\exp \left(wx\right)}{1+\exp \left(wx\right)},P(y=0|x)=\frac{1}{1+\exp \left(wx\right)};$

Wherein, user's trip data that x represents, y=1 represents and is predicted as the male sex, and y=0 represents and is predicted as women, and w represents coefficient.

S4: according to described two classification forecast models, and the trip data of terminal, predicts the sex of user corresponding to terminal.

Specifically, according to the two classification forecast models obtained, and the trip data of any terminal, can the user sex corresponding to terminal predict.

Present embodiments provide a kind of sex and know method for distinguishing, identify according to the sex of speech data to user corresponding to part terminal, trip data training according to this part terminal obtains two disaggregated models, does not have the sex of user corresponding to the terminal of speech data to identify to other.Integrated voice of the present invention, trip data and friend relation determine the sex of user corresponding to terminal, analyze comprehensively, and accuracy are higher.Thus further according to the preference of different sexes, the user for different sexes pushes different personalized products or application.

In the present embodiment, according to described speech data in step S1, the sex of user corresponding to described multiple terminal is identified, specifically comprises:

S11: if the speech data of a terminal comprises a kind of speech data of sex, then using the sex of this sex as user corresponding to described terminal;

S12: if the speech data of a terminal comprises the speech data of two kinds of sexes, then filter out the speech data that this terminal is corresponding.

As can be seen here, for the terminal comprising two kinds of sex speech datas, the sex of user cannot be confirmed exactly, therefore this partial data is removed, to improve the prediction accuracy of the follow-up two classification forecast models obtained.

In the present embodiment, after step S1, the method also comprises the steps:

A01, obtain and identified user's friend relation corresponding to the terminal of user's sex.

Specifically, the detailed process obtaining user's friend relation can be: for each terminal, obtain and wherein can record and get record sharing of sharing information, can record and get record sharing of sharing information according to each, by share described can sharing information first terminal and can the second terminal of sharing information associate described in getting, generate user's friend relation that multiple terminal is corresponding.

A02, according to described friend relation, to obtain and described user good friend and the highest first user of pre-set priority each other.

Specifically, pre-set priority the highest i.e. both expressions cohesion is the highest, and both interactions are the most frequent.

A03, judge described first user and the described user having identified sex different in nature each other.

As can be seen here, the present embodiment also can according to the user's buddy list of terminal identifying user's sex, obtains good friend each other and the most close user of friend relation, this user is regarded as the above-mentioned spouse or the male/female friend that have identified sex user.

As shown in Figure 2, still another embodiment provides a kind of structural representation of device of sex identification for the disclosure, this device comprises: the first sex recognition unit 201, acquiring unit 202, forecast model set up unit 203 and gender prediction's unit 204.Wherein:

First sex recognition unit 201, for obtaining the speech data of multiple terminal in the first preset time period, and according to described speech data, identifies the sex of user corresponding to described multiple terminal;

Acquiring unit 202, for obtaining in the second preset time period the trip data of the multiple terminals having identified user's sex;

Forecast model sets up unit 203, for the trip data of described multiple terminal is carried out two classification based trainings as corpus, obtains the two classification forecast models for gender prediction;

Gender prediction's unit 204, for according to described two classification forecast models, and the trip data of terminal, the user sex corresponding to terminal is predicted.

In the present embodiment, the first sex recognition unit 201, for:

If the speech data of a terminal comprises a kind of speech data of sex, then using the sex of described sex as user corresponding to terminal;

If the speech data of a terminal comprises the speech data of two kinds of sexes, then using the sex of sex corresponding for a fairly large number of voice as user corresponding to terminal.

In the present embodiment, for the system of calling a taxi, then above-mentioned trip data can be to go out line frequency, travel time section distribution, the POI distribution of trip purpose ground and frequency, user integral store redemption data week, give tip amount, give tip frequency, by certificate ratio and with one or more in certificate mean value.

In the present embodiment, described device also comprises second sex recognition unit, for:

Obtain and identified user's friend relation corresponding to the terminal of user's sex;

According to described friend relation, obtain and described user good friend and the highest first user of pre-set priority each other;

Judge described first user and the described user having identified sex different in nature each other.

In the present embodiment, the described two classification forecast models for sex test are:

$\underset{i}{\Σ}\[y_{i}LogP(y=1|x)+(1-y_i)LogP(y=0|x)\]$

Wherein, $P(y=1|x)=\frac{\exp \left(wx\right)}{1+\exp \left(wx\right)},P(y=0|x)=\frac{1}{1+\exp \left(wx\right)};$

Wherein, user's trip data that x represents, y=1 represents and is predicted as the male sex, and y=0 represents and is predicted as women, and w represents coefficient.。

For device embodiment, due to itself and embodiment of the method basic simlarity, so description is fairly simple, relevant part illustrates see the part of embodiment of the method.

Should be noted that, in all parts of system of the present disclosure, the function that will realize according to it and logical partitioning has been carried out to parts wherein, but, the disclosure is not limited to this, can repartition all parts as required or combine, such as, can be single parts by some component combinations, or some parts can be decomposed into more subassembly further.

All parts embodiment of the present disclosure with hardware implementing, or can realize with the software module run on one or more processor, or realizes with their combination.It will be understood by those of skill in the art that the some or all functions that microprocessor or digital signal processor (DSP) can be used in practice to realize according to the some or all parts in the system of disclosure embodiment.The disclosure can also be embodied as part or all equipment for performing method as described herein or device program (such as, computer program and computer program).Realizing program of the present disclosure and can store on a computer-readable medium like this, or the form of one or more signal can be had.Such signal can be downloaded from internet website and obtain, or provides on carrier signal, or provides with any other form.

It should be noted that above-described embodiment is described the disclosure instead of limits the disclosure, and those skilled in the art can design alternative embodiment when not departing from the scope of claims.In the claims, any reference symbol between bracket should be configured to limitations on claims.Word " comprises " not to be got rid of existence and does not arrange element in the claims or step.Word "a" or "an" before being positioned at element is not got rid of and be there is multiple such element.The disclosure can by means of including the hardware of some different elements and realizing by means of the computing machine of suitably programming.In the unit claim listing some devices, several in these devices can be carry out imbody by same hardware branch.Word first, second and third-class use do not represent any order.Can be title by these word explanations.

Above embodiment is only suitable for the disclosure is described; and not to restriction of the present disclosure; the those of ordinary skill of relevant technical field; when not departing from spirit and scope of the present disclosure; can also make a variety of changes and modification; therefore all equivalent technical schemes also belong to category of the present disclosure, and scope of patent protection of the present disclosure should be defined by the claims.

Claims (10)

1. sex knows a method for distinguishing, and it is characterized in that, the method comprises:

Obtain the speech data of multiple terminal in the first preset time period, and according to described speech data, the sex of user corresponding to described multiple terminal is identified;

Obtain in the second preset time period the trip data of the multiple terminals having identified user's sex;

The trip data of described multiple terminal is carried out two classification based trainings as corpus, obtains the two classification forecast models for gender prediction;

According to described two classification forecast models, and the trip data of terminal, predicts the sex of user corresponding to terminal.

2. method according to claim 1, is characterized in that, described according to described speech data, identifies, comprising the sex of user corresponding to described multiple terminal:

If the speech data of a terminal comprises a kind of speech data of sex, then using the sex of this sex as user corresponding to described terminal;

If the speech data of a terminal comprises the speech data of two kinds of sexes, then filter out the speech data that this terminal is corresponding.

3. method according to claim 1, it is characterized in that, described trip data is go out line frequency, travel time section distribution, the information point POI distribution of trip purpose ground and frequency, user integral store redemption data week, give tip amount, give tip frequency, by certificate ratio and with one or more in certificate mean value.

4. method according to claim 1, is characterized in that, described according to described speech data, after identifying, also comprises the sex of user corresponding to described multiple terminal:

Obtain and identified user's friend relation corresponding to the terminal of user's sex;

According to described friend relation, obtain and described user good friend and the highest first user of pre-set priority each other;

Judge described first user and the described user having identified sex different in nature each other.

5. method according to claim 1, is characterized in that, described in obtain for gender prediction two classification forecast models be:

$\underset{i}{\Σ}\[y_{i}LogP(y=1|x)+(1-y_i)LogP(y=0|x)\]$

Wherein, $P(y=1|x)=\frac{\exp \left(wx\right)}{1+\exp \left(wx\right)},P(y=0|x)=\frac{1}{1+\exp \left(wx\right)};$

Wherein, user's trip data that x represents, y=1 represents and is predicted as the male sex, and y=0 represents and is predicted as women, and w represents coefficient.

6. a device for sex identification, is characterized in that, this device comprises:

First sex recognition unit, for obtaining the speech data of multiple terminal in the first preset time period, and according to described speech data, identifies the sex of user corresponding to described multiple terminal;

Acquiring unit, for obtaining in the second preset time period the trip data of the multiple terminals having identified user's sex;

Forecast model sets up unit, for the trip data of described multiple terminal is carried out two classification based trainings as corpus, obtains the two classification forecast models for gender prediction;

Gender prediction's unit, for according to described two classification forecast models, and the trip data of terminal, predicts the sex of user corresponding to terminal.

7. device according to claim 6, is characterized in that, described first sex recognition unit, for:

If the speech data of a terminal comprises a kind of speech data of sex, then using the sex of this sex as user corresponding to described terminal;

If the speech data of a terminal comprises the speech data of two kinds of sexes, then filter out the speech data that this terminal is corresponding.

8. device according to claim 6, it is characterized in that, described trip data is go out line frequency, travel time section distribution, the information point POI distribution of trip purpose ground and frequency, user integral store redemption data week, give tip amount, give tip frequency, by certificate ratio and with one or more in certificate mean value.

9. device according to claim 6, is characterized in that, described device also comprises second sex recognition unit, for:

Obtain and identified user's friend relation corresponding to the terminal of user's sex;

According to described friend relation, obtain and described user good friend and the highest first user of pre-set priority each other;

Judge described first user and the described user having identified sex different in nature each other.

10. device according to claim 6, is characterized in that, the described two classification forecast models for sex test are:

$\underset{i}{\Σ}\[y_{i}LogP(y=1|x)+(1-y_i)LogP(y=0|x)\]$

Wherein, $P(y=1|x)=\frac{\exp \left(wx\right)}{1+\exp \left(wx\right)},P(y=0|x)=\frac{1}{1+\exp \left(wx\right)};$

Wherein, user's trip data that x represents, y=1 represents and is predicted as the male sex, and y=0 represents and is predicted as women, and w represents coefficient.