CN106708939A - Target person scoring and pushing methods, apparatuses and systems - Google Patents

Abstract

The technical scheme of the invention provides target person scoring and pushing methods, apparatuses and systems. The target person scoring method comprises the steps of obtaining a plurality of first scoring indexes and distribution of dimensions under the first scoring indexes; performing statistics on probability distribution of the dimensions under the first scoring indexes for a target person; calculating the fit degrees of the target person for the first scoring indexes based on state distribution of the dimensions under the first scoring indexes; taking an average value for the fit degrees of the first scoring indexes belonging to the same index type to obtain index values; and outputting a score of the target person based on a weighted average result of all the index values and corresponding index weights. According to the technical scheme, automatic evaluation, selection and recommendation of the target person can be realized.

Description

Method, device and system for grading and pushing target person

Technical Field

The invention belongs to the technical field of computers, and particularly relates to a method for grading a target figure, a method for pushing the figure, a device for grading the target figure and a system for pushing the figure.

Background

With the popularization of the internet, the same website can be used by all kinds of people. For example, a navigation web page is used by people with the age span of more than 60 years, and their occupation varies, and the demand for internet surfing varies greatly. Massive information is brought forward like tide, and people have already entered an information explosion era, on the background, on one hand, users are not easy to find interesting contents from the information, and on the other hand, a large amount of information is unobtrusive and cannot be obtained by the users.

How to realize the pairing between the information and the corresponding user is that the user can obtain more and more practical information, the attribute of the target user needs to be judged, and based on the attribute, the condition, the need and the hobbies of the target user are known, and then the relevant information can be provided for the target user.

At the present stage, there is no very good way to solve the above mentioned problems.

Disclosure of Invention

The technical problem solved by the technical scheme of the invention is as follows: on the basis of processing the information data of the target user, scoring is carried out through an algorithm, and then recommendation of corresponding information is achieved.

In order to solve the above technical problem, the technical solution of the present invention provides a method for scoring a target person, including:

acquiring a plurality of first scoring indexes and state distribution of dimensionality under the first scoring indexes;

counting the state distribution of each dimension of the first scoring index aiming at the target person;

calculating the degree of engagement of the target person on the first scoring index based on the state distribution of each dimension of the first scoring index;

averaging the fitness of the first grading indexes belonging to the same index type to obtain index values;

and outputting the score of the target person based on the weighted average result of all the index values and the corresponding index weights.

Optionally, the index types include a first index type and a second index type, and the method for scoring the target person further includes:

obtaining the attention degree of the target person;

calculating an attention index multiplier of the target person based on the attention of the target person, the attention index multiplier being the attention of the target person relative to other persons;

if the first scoring index belongs to a first index type, the step of averaging the fitness of the first scoring indexes belonging to the same index type to obtain the index value comprises the following steps: multiplying the average value of the degree of engagement of the first grading index and a multiplier of an attention index of the target person to obtain the index value;

if the first scoring index belongs to the second index type, the step of averaging the fitness of the first scoring indexes belonging to the same index type to obtain the index value comprises the following steps: and taking the average value of the fitness of the first scoring index as the index value.

Optionally, the method further includes:

acquiring a plurality of second grading indexes;

counting the statistical value of the second scoring index for the target person;

calculating a relative liveness of the second scoring index based on the statistical value of the second scoring index;

and taking the average value of the relative liveness of the second scoring index to obtain the index value.

Optionally, for the second scoring index, the index value is obtained based on the following steps:

acquiring the activity of the target character;

calculating an activity multiplier for the target character based on the activity of the target character;

and multiplying the average value of the relative liveness of the second grading index and the liveness multiplier of the target character to obtain the index value.

Optionally, the engagement degree is cos θ_iThe method comprises the following steps:

wherein i is the index number, R_iA state distribution matrix for each dimension of the first scoring index,a state distribution matrix (R) for each index of the target person_i)^TIs R_iThe transposed matrix of (2).

Optionally, let the attention index multiplier be T_iThe method comprises the following steps:

wherein i is the number of the target person, the attention of the target person, and t₁，t₂，...，t_nAnd n is the number of the other persons and the number of the target persons.

Optionally, setting the relative activity of the second scoring index as F_iThe method comprises the following steps:

wherein i is the number of the target person, the statistical value of the second score index of the target person, the statistical value of the second score indexes of the other persons and the target person, and n is the number of the other persons and the target person.

Optionally, set the liveness multiplier to S_iThe method comprises the following steps:

wherein i is the number of the target person, g_iIs the liveness of the target character, g₁，g₂，...，g_nAnd n is the number of the other characters and the target characters.

Optionally, the outputting the score of the target person based on the multiplication result of all the index values and the corresponding index weights thereof includes:

and accumulating the multiplication results to obtain the score of the target person.

Optionally, the outputting the score of the target person based on the multiplication result of all the index values and the corresponding index weights thereof includes:

acquiring a third scoring index;

counting the third scoring index value for the target person;

and accumulating the multiplication results and multiplying the accumulated value and the third score index numerical value to obtain the score of the target person.

In order to solve the technical problem, the technical solution of the present invention further provides a method for pushing a person, including:

scoring all target persons based on the method;

and pushing the target person based on the score of the target person.

Optionally, the method further includes:

receiving a user character pushing request;

the scoring all target persons is performed based on the request.

In order to solve the above technical problem, the present invention further provides a device for scoring a target person, including:

the acquisition unit is suitable for acquiring a plurality of first scoring indexes and the distribution of dimensionality under the first scoring indexes;

the statistic unit is suitable for counting the state distribution of each dimension of the first grading index aiming at the target person;

the calculating unit is suitable for calculating the engagement degree of the target person on the first scoring index based on the state distribution of each dimension of the first scoring index;

the averaging unit is suitable for averaging the fitness of the first grading indexes belonging to the same index type to obtain an index value;

and the output unit is suitable for outputting the score of the target person based on the multiplication result of all the index values and the corresponding index weights.

In order to solve the above technical problem, a technical solution of the present invention provides a system for pushing a person, including:

the device is suitable for outputting all the scores of the target characters;

and the pushing unit is suitable for pushing the target person based on the score of the target person.

The technical scheme of the invention at least comprises the following beneficial effects:

the technical scheme of the invention is based on a preset evaluation system, the evaluation system is provided with a tag set, the tag set can be related to evaluation classification (games, shopping, economy and the like), tag evaluation is carried out based on webpage data browsed by a browser user, and the attribute of gender of male and female is judged by making preference according to the data of a big data/history browser user so as to better push the webpage. This helps to achieve that the pushed information meets the requirements of the browser user.

On the other hand, the technical scheme of the invention can evaluate various public characters, judge the attributes of the public characters, determine various objective evaluation information of the comprehensive public characters, and evaluate the public characters based on the evaluation information, thereby adaptively recommending and selecting the characters, realizing the comprehensive evaluation of various character information under the network and big data background, and realizing the automatic selection and recommendation of the target characters.

Drawings

Other features, objects and advantages of the invention will become more apparent upon reading of the detailed description of non-limiting embodiments with reference to the following drawings:

fig. 1 is a flowchart illustrating a method for scoring a target person according to a first embodiment of the present invention;

fig. 2 is a flowchart of a method for scoring a target person according to a second embodiment of the present invention;

fig. 3 is a flowchart of a method for scoring a target person according to a third embodiment of the present invention;

FIG. 4 is a flowchart of a method for obtaining an indicator value according to a variation of the third embodiment of the present invention;

fig. 5 shows a flowchart of a method for pushing a person according to a fourth embodiment of the present invention;

fig. 6 is a block diagram illustrating an apparatus for scoring a target person according to a fifth embodiment of the present invention.

Detailed Description

In order to better and clearly show the technical scheme of the invention, the invention is further described with reference to the attached drawings.

It should be understood that the present invention is mainly applicable to, but not limited to, such a scenario, and those skilled in the art understand that such a control process is of great significance, and a browser user may correspond to multiple users; one user may also be multiple browser users. And (3) setting preferences through data of a big data/history browser user, and judging the attributes of gender of men and women so as to better push the webpage. The technical scheme of the invention is that an evaluation system is set, the evaluation system is provided with a label set, the label set is relevant to evaluation classification (games, shopping, economy and the like), label evaluation is carried out based on webpage data browsed by a browser user, and calculation on a basic score (0) is carried out on the corresponding evaluation classification in evaluation, such as (+1) for male and (-1) for female; and adding the final calculated values of all the evaluation classifications to obtain a negative value or a positive value on the basic score, thereby recommending the webpage/product.

Fig. 1 is a flowchart illustrating a method for scoring a target person according to a first embodiment of the present invention. The method comprises the following specific steps:

step S101 is entered, and a plurality of first scoring indexes and state distribution of dimensionality under the first scoring indexes are obtained. Specifically, the obtaining is to detect whether a user input instruction exists on a page of the user terminal and to determine the type of the user input instruction. The user terminal may be a browser in a computer, a mobile phone, etc. The page of the user terminal is usually one page in website management service accessed through a browser, and the server sends back response information according to the operation of the user on the accessed page; the first score refers to a set of tags in the rating system that are relevant to the rating classification (game, shopping, economy, etc.).

Step S102 is executed, specifically, for the target person, statistics is performed on state distribution of each dimension of the first score index.

And step S103, calculating the engagement degree of the target person on the first scoring index based on the state distribution of each dimension of the first scoring index. Specifically, the degree of engagement is the degree of similarity between the target person and the first score index determined by the system. More specifically, let the engagement degree be cos θ_iThe method comprises the following steps:

wherein i is the index number, R_iA state distribution matrix for each dimension of the first scoring index,a state distribution matrix (R) for each index of the target person_i)^TIs R_iThe transposed matrix of (2).

Step S104 is executed, and the first scoring index integrating degrees belonging to the same index type are averaged to obtain an index value. Specifically, the index value is a determination of the attribute of the target person, and may be used to indicate that the target person is more male/female, for example.

And S105, outputting the score of the target person based on the weighted average result of all the index values and the corresponding index weights. Specifically, the score is a comprehensive score, that is, an attribute of the target person reflected by the target person indexes in a comprehensive manner.

Further, the outputting the score of the target person based on the multiplication result of all the index values and the corresponding index weights thereof includes: and accumulating the multiplication results to obtain the score of the target person.

Fig. 2 is a flowchart illustrating a method for scoring a target person according to a second embodiment of the present invention. The index types comprise a first index type and a second index type, wherein the second index type belongs to the index types, the index types are divided, one part of the index types (corresponding to the first index type) are subjected to average value multiplication processing with attention, and the other part of the index types (corresponding to the second index type) are subjected to average value processing. The method comprises the following specific steps:

the process proceeds to step S201, and the attention of the target person is acquired. Specifically, the target person is not a real person, but refers to an internet entrance such as a browser; the attention may be click rate or exposure.

Step S202 is performed to calculate an attention index multiplier of the target person based on the attention of the target person. Specifically, the attention index multiplier is the attention of the target person relative to other persons. More specifically, let the attention index multiplier be T_iThe method comprises the following steps:

wherein i is the number of the target person, t_iIs the attention of the target person, t₁，t₂，...，t_nAnd n is the number of the other persons and the number of the target persons.

And step S203, judging the type of the first scoring index.

Executing step S204, if the first score indicator belongs to the first indicator type, the averaging the fitness of the first score indicators belonging to the same indicator type to obtain the indicator value includes: multiplying the average value of the degree of engagement of the first grading index and a multiplier of an attention index of the target person to obtain the index value; if the first scoring index belongs to the second index type, the step of averaging the fitness of the first scoring indexes belonging to the same index type to obtain the index value comprises the following steps: and taking the average value of the fitness of the first scoring index as the index value.

Fig. 3 is a flowchart illustrating a method for scoring a target person according to a third embodiment of the present invention.

And step S301 is entered to acquire a plurality of second scoring indexes. Specifically, the second score and the first score are all a certain category of the score, including gender, occupation, and the like, and are only distinguished when expressed.

Step S302 is executed to count the statistical value of the second score index for the target person.

Step S303 is entered, and the relative activity of the second scoring index is calculated based on the statistic value of the second scoring index. Specifically, the relative activity of the second scoring index is F_iThe method comprises the following steps:

wherein i is the number of the target person, the statistical value of the second score index of the target person, the statistical value of the second score indexes of the other persons and the target person, and n is the number of the other persons and the target person.

And executing step S304, and taking the average value of the relative liveness of the second scoring index to obtain the index value.

Further, the index value is obtained based on the method shown in fig. 4. The method comprises the following specific steps;

and step S401 is carried out, and the activity of the target person is obtained.

Step S402 is performed to calculate an activity multiplier for the target character based on the activity of the target character. Specifically, let the liveness multiplier be S_iThe method comprises the following steps:

wherein i is the number of the target person, g_iIs the liveness of the target character, g₁，g₂，...，g_nAnd n is the number of the other characters and the target characters.

And S403, multiplying the average value of the relative liveness of the second scoring index by the multiplier of the liveness of the target person to obtain the index value.

Further, the outputting the score of the target person based on the multiplication result of all the index values and the corresponding index weights thereof includes: acquiring a third scoring index; counting the third scoring index value for the target person; and accumulating the multiplication results and multiplying the accumulated value and the third score index numerical value to obtain the score of the target person.

Fig. 5 shows a flowchart of a method for pushing a person according to a fourth embodiment of the present invention. The fourth embodiment scores all the target persons based on any one of the first, second and third embodiments, and then pushes the scores of the target persons to perform task pushing. The method comprises the following specific steps:

the process advances to step S501, and a user character push request is received. Specifically, the user character is an operator at a browser end; the push request refers to a requirement issued by the user character during browser operation, and the requirement may be directly conveyed by the user character or implicitly required in some operation. For example, a user searching for a pen at night, the implicit requirement may be a copybook, ink, etc.

Step S502 is performed, and the scoring of all the target persons is performed based on the request.

Fig. 6 is a block diagram of an apparatus for scoring a target person according to a fifth embodiment of the present invention, which includes an obtaining unit 61, a counting unit 62, a calculating unit 63, an averaging unit 64, and an output unit 65. The obtaining unit 61 is adapted to obtain a plurality of first scoring indexes and a distribution of dimensions under the first scoring indexes; the statistic unit 62 is adapted to count the state distribution of each dimension of the first scoring index for the target person; the calculating unit 63 is adapted to calculate the degree of engagement of the target person with respect to the first scoring index based on the state distribution of each dimension of the first scoring index; the averaging unit 64 is adapted to average the fitness of the first scoring indicators belonging to the same indicator type to obtain an indicator value; the output unit 65 is adapted to output the score of the target person based on the multiplication result of all index values by their corresponding index weights.

It should be noted that the structure of the device for scoring the target person is not limited to this embodiment, for example, the acquiring unit 61, the counting unit 62, the calculating unit 63, and the averaging unit 64 may be combined into a same module, such as a processing module, and the processing module and the output unit 65 complete the scoring function; for example, the averaging means 64 may be combined with the calculating means 63, or the counting means 62 and the averaging means 64 may be combined with the calculating means 63. The technical scheme of the invention does not limit the structural form.

Further, the device for scoring the target person and the pushing unit jointly form a system for pushing the task. The pushing unit is suitable for pushing people based on the scores of the target people.

According to the technical features of the technical solution of the present invention, this embodiment further provides an application example, that is, the actor is scored by using the method for scoring the target character, and a scoring model of the actor is established based on the method for scoring the target character. When the actor scoring model is established by using the method for scoring the target character, the purpose of the method comprises the following steps: the influence of actor factors on the movie and television play benefits is manifold and is reflected in the aspects of popularity, skills, liveness and the like; selecting actors requires multi-dimensional consideration to maximize benefit. Due to the fact that the dimensionalities involved in actor selection are large, and the difference between the dimensionalities is large, complete consideration is difficult to form by means of subjective judgment. The actor scoring model is established to ensure the comprehensiveness and diversity of evaluation indexes, and different indexes are reasonably integrated through certain statistics, so that the readability and the practicability of results are enhanced. In addition, the factors influencing the performance of the movie and television play are considered to be various, and besides the factors of actors, the factors of the play, the shooting, the propaganda and the like are also considered; therefore, the actor scoring model starts from one of the links, and the actor with the maximum benefit is selected under the condition of giving the script and not considering shooting and promotion. Among these conditions, a given scenario is the most important prerequisite, i.e., the subject, genre, and scenario to be captured by a movie or television series can be sufficiently interpreted.

Based on the above thought, in combination with the method for scoring a target character described in this embodiment, when the above technical solution is applied to actor scoring, first, the method includes a step of setting the following index system, which is detailed in a set actor scoring model index.

Table one: actor score model index

Based on table one, the present application example divides the actor scoring model into the following index types, including: script-vermicelli conformity (C), vermicelli multiplier, script-ability conformity (R), popularity (F) and liveness multiplier. Wherein:

the script-fan integrating degree (C) indicated by the first table comprises the following index dimensions (namely index names):

(1) age group (C)₁)

The age groups are divided into 6 segments, 0-12 represent juvenile population, 12-18 represent adolescent population (high school and middle school), 18-24 represent college student population, 24-35 represent young people who enter the workplace at the beginning, 35-55 represent middle school population, and more than 55 represent elderly population.

The ages of the actor fans are distributed in frequency, and for the actors of which the ages of the fans are younger, the fans of the actors are mainly students, and a typical distribution can be as follows: c₁The distribution of the ages of the actor fans may be different according to the distribution of the ages of the fan crowd, and for the actors with the aged fan fans, the distribution may be C₁＝(0，0.2，0.3，0.4，0.1，0)。

(2) Sex (C)₂)

Based on the female proportion of the vermicelli, the vermicelli is divided into 6 grades, namely 0-15%, 15-30%, 30-50%, 50-70%, 70-85% and more than 85%.

The sex index is a state, and if the female proportion of the actor fan is 60%, the distribution state is: c₂If the female proportion of the actor fan is 44%, the distribution state is (0, 0, 0, 1, 0, 0): c₂＝(0，0，1，0，0，0)

(3) Tidal current State (C)₃)

The trend state reflects the attention of the actor to the playpen. Investigating 50-100 active fans, the average proportion of 20 microblogs which are closest to the fans of entertainment; the proportion is divided into 6 grades, namely 0-15%, 15-30%, 30-50%, 50-70%, 70-85% and more than 85%.

The power flow state index is alsoIn one state, if the average percentage of the microblogs in the entertainment circle is 34%, then: c₃＝(0，0，1，0，0，0)

(4) Hobby (C)₄)

The theme and plot of a script often can correspond to certain life states, such as home, nature, sports, reading, doing exercises, etc., and different people may select different scripts. The interest and hobby index classifies the state of life into 6 categories, static (house, art, workplace), dynamic (sports, nature), and ideological (fantasy).

The interest and hobby index is also a frequency distribution, and 50-100 active fans are investigated to obtain the proportion of various living states. If the ratio is 35%, 20%, 10%, 5%, 25%, respectively: c₄＝(0.35，0.2，0.1，0.05，0.05，0.25)

And the second table shows the conformity (C) of the script-vermicelli in each index and the dimension distribution thereof.

Table two: theater-vermicelli conformity (C) degree index and dimensionality

Based on the above table one and table two, a calculation method for calculating the script-fan engagement degree (C) is given below.

According to the target population characteristics of the script-fan conformity (C), if a script is given, the characteristics of the target population can be expressed as follows according to 4 dimensions of age bracket, gender, trend state and hobbies:

wherein,andrepresenting age and hobby, is a probability distribution, andandrepresenting gender and tidal current status, is a condition.

When the fitness is calculated, the cosine similarity is used, and the fitness of some index (age, sex, trend state and interest) is calculated as follows:

the script-fan engagement (C) of the actor is the average of 4 index engagements:

the scenario-capability engagement (R) indicated in table one includes the following index dimensions (i.e., the index name):

(1) the role acting force (R)₁)

Actor's interpretation of the character will affect the overall quality of the play, and the actor's experience will reflect the competency of the character in the play. According to the role types, the indexes are divided into 7 dimensions of sadness, happiness, emotion, coldness, stability, strength and sex.

The character acting force (R1) is a frequency distribution, and the 7-dimension ratio can be calculated by counting all the character types played by the actor.

(2) The scenario plays a role (R)₂)

The type of the movie and television play played by the actor also influences the interpretation of the play, and similar to the role playing force, indexes are divided into 7 dimensions of ancient costume, history, city, idol, science fiction, suspicion and mysterious illusion according to the type of the movie and television play.

The index is also a frequency distribution, counts the types of all plays performed by the actors, and calculates the ratio of each dimension. Table three gives the transcript-ability fit (R) in each index and its dimensional distribution.

Table three: script-ability engagement (R) index and dimensionality

Based on the above table one and table two, the following calculations are given: script-ability integrating degree (R) index and dimension calculation method.

From a given transcript, the role-playing force and the storyline-playing force may be expressed as follows from the dimension distribution of table three:

whereinAccording to the performance script before the actor is counted, the probability distribution of the actor on the types of sadness, happiness, sentiment, coldness, stability, strength and sex of the role of the script is calculated,is the probability distribution of actors on the script story type antique, history, city, idol, science fiction, suspicion, and fantasy.

It should be noted that, in the following description,andthe type of the representative character and the type of the drama in the present application case are one probability distribution, but are closer to the state distribution.

Similar to the calculation of the script-fan integrating degree (C), the script-capability integrating degree (R) is based on two indexes of role acting force and scenario acting force, and the calculation formula of the script-capability integrating degree (R) comprises cosine calculation:

the comprehensive indexes of script-ability integrating degree (R) are as follows:

for the first named degree (F) of the table, the calculation process comprises the following steps:

setting a relative scoring system, and combining the index names of the table I, wherein the known degree index comprises:

F₁: year of the lane-the time from the first game play to the scoring, in years;

F₂: showing times-the number of movies and television shows which are shown together when the score is up;

F₃: obtaining the number of times of the international prize term when the evaluation is finished;

F₄: public rating-average rating of all movie shows on video websites by score.

If Y is₁₁，Y₁₂，...，Y_1nIs the date of the N actors, F of the ith actor₁The method comprises the following steps:

f of the ith actor is calculated in the same manner₂、F₃And F₄Score, denoted as F_2i，F_3iAnd F_4i。

The overall scores of the awareness (i.e., relative liveness in the above embodiment) of actor i are:

wherein j is 1-4.

The calculation of the liveness multiplier (S) indicated in Table one includes: the popularity represents the performance achievement of the actor and is a static historical index. However, the decay rate of the actor's attention is extremely fast, and therefore, recent liveness has a large effect on awareness.

Let gi be the search index of the ith actor in a certain period (for example, the last 1 year or several years), the search target may be based on an average of one or more statistical results in all search engines such as a hundred-degree website, the statistical result may be the sum of the search times of the ith actor on the search engine, or a search ratio calculated based on the sum, and since the statistical result may be objectively evaluated according to an algorithm or statistics of the search statistics, the present application does not limit the specific definition of the statistical result.

Based on the above g_iThe liveness multiplier (S) of the ith actor is:

the actor liveness multiplier (S) is used in the actor rating model of the present application to adjust the awareness (F) of the actor, although it is used aloneThe actor scoring model can be established by using the popularity (F) algorithm of the application example, but the accuracy is poor. The application example also uses the activeness multiplier (S) of the actor to adjust the popularity (F) of the actor so as to enable the index in the final actor scoring model to be more accurate, and based on the explanation, the adjusted popularity (F) of the actor is provided_i′＝F_i·S_i。

And by continuously combining the index parameters in the table I, the script-fan conformity (C) measures the relative benefits of the actors in the aspect of correlation, but the fan scale can also embody the scale benefits. The truly active fans can be reflected in actual actions, so that at least one product statistic of the sales quantity, the shelf quantity and the like of the products such as the same commodities of the actors or other works related to the actors displayed in a network seller or other statistic platforms can be used as an index calculation basis.

If the number of the same-style commodities of the actors on the Taobao is used as an index calculation basis, let ti be the number of the same-style commodities of the ith actor on the Taobao, let 1.

The fan multiplier (T) is used for adjusting the script-fan integrating degree (C) in the actor scoring model, so that the adjusted script-fan integrating degree (C) can be used for evaluating the objectivity of the actor fan effect. Script-vermicelli conformity degree C of ith actor adjusted by vermicelli multiplier (T)'_i＝C_i·T_i。

Of course, it should be noted that in other embodiments, it is also possible to directly output the script-fan engagement (C) without adjusting the fan multiplier (T).

After the initial establishment of the model parameters, the adjusted script-fan conformity, script-ability conformity and the adjusted popularity of the actor constitute an evaluation model of the actor, and an evaluation model P is (C ', R, F'). That is, the evaluation model in the present application example includes three parameters, i.e., the adjusted scenario-fan conformity, the scenario-ability conformity, and the adjusted popularity.

Each parameter in the above evaluation model, i.e. adjusted transcript-fan conformity, transcript-ability conformity and adjusted popularity, also has its own output driving weight. According to different investments, the output driving weight of the movie and television play is different, that is, according to different driving forces of actors, the actors are subjected to propaganda driving, self performance strength and the like when facing the public, and each parameter of the actor evaluation model can be compared with the propaganda driving and the performance quality driving to set the weight value between each parameter.

For example, the promotion driver realizes the profit depending on the popularity and the channel under the condition of large cost space, so the tradeoffs among the fan conformity, the capability conformity and the popularity are different, and the influence on the popularity is large in general, for example, the weighted values among the script-fan conformity, the script-capability conformity and the popularity can be set to be (0.2, 0.2, 0.6). In addition, quality drives have gained public acceptance by virtue of good performance, with an expansion in revenue through public praise. In this case, the actor's personality is relatively more important, so the trade-off of fan-goodness, ability-goodness, and popularity may be (0.3, 0.5, 0.2). The division of the weighted values is only used as an example, and a more appropriate parameter weight is set for each parameter in the evaluation model according to a set driving factor (such as the promotion driver, the actor instance, etc.), so as to accurately infer the balance between each parameter in the current stage model, which can be determined by a person skilled in the art according to the technical feature flow disclosed in the technical solution of the present invention.

Setting the weight of each parameter in the evaluation model based on the driving factors according to the evaluation model, and setting the engagement degree of script-bean vermicelli, the engagement degree of script-ability and the adjustment degreeThe weight W of the whole unknown degree is equal to (W)_c’，W_R，W_F’) Wherein W is_c’、W_RAnd W_F’The adjusted script-fan integrating degree weight, the script-ability integrating degree weight and the adjusted popularity weight are respectively.

For this application example, the final scoring model of the actor is: score ═ P^T·W。

In a variation of this application, the risk value of the actor may be comprehensively evaluated by being included in the actor evaluation model. The actors are public characters, and the public image and public opinion guidance of the actors have great influence on input and output. Positive, good appearance is a guarantee of revenue realization, whereas negative messages are the most important for the potential impact of revenue. Score P of the above scoring model^TOn the basis of W, negative messages can be incorporated into a scoring model, news disclosed on the network or media in an appointed period (such as 6 months) can be searched, a certain network platform can be determined to conduct news search, such as hundredths or surf, and the risk value Z of the actor can be judged. The specific value of the risk value Z may be determined by any one of the following methods:

(1) if serious negative messages such as violence, law violation and crime occur, Z is 0;

(2) general negative messages (life, emotion, conflict, etc.), if the news accounts for more than 50%, Z is 0.5;

(3) generally, if the negative message ratio is less than 50%, Z is 0.8;

(4) without a negative message, Z is 1.

Therefore, the actor assessment model after adding the risk value adjustment specifically includes:

Adjusted Score＝P^T·W·Z。

the above application examples specifically illustrate how the method of scoring a target character is applied to scoring an actor, and of course, the scoring object of the above method is not limited to actors, and in fact, evaluation of various public characters is possible. The technical scheme of the invention can evaluate various public figures, judge the attributes of the public figures to determine various objective evaluation information of the comprehensive public figures, and evaluate the public figures based on the evaluation information, thereby adaptively recommending and selecting the figures, realizing the comprehensive evaluation of various figure information under the network and big data background and realizing the automatic selection and recommendation of target figures.

The foregoing describes specific embodiments of the present invention. It is to be understood that the present invention is not limited to the specific embodiments described above, and that various changes and modifications may be made by one skilled in the art within the scope of the appended claims without departing from the spirit of the invention.

Claims (14)

1. A method of scoring a target person, comprising:

acquiring a plurality of first scoring indexes and state distribution of dimensionality under the first scoring indexes;

counting the state distribution of each dimension of the first scoring index aiming at the target person;

calculating the degree of engagement of the target person on the first scoring index based on the state distribution of each dimension of the first scoring index;

averaging the fitness of the first grading indexes belonging to the same index type to obtain index values;

and outputting the score of the target person based on the weighted average result of all the index values and the corresponding index weights.

2. The method of scoring a target person of claim 1, wherein the indicator types comprise a first indicator type and a second indicator type, the method further comprising:

obtaining the attention degree of the target person;

calculating an attention index multiplier of the target person based on the attention of the target person, the attention index multiplier being the attention of the target person relative to other persons;

if the first scoring index belongs to a first index type, the step of averaging the fitness of the first scoring indexes belonging to the same index type to obtain the index value comprises the following steps: multiplying the average value of the degree of engagement of the first grading index and a multiplier of an attention index of the target person to obtain the index value;

if the first scoring index belongs to the second index type, the step of averaging the fitness of the first scoring indexes belonging to the same index type to obtain the index value comprises the following steps: and taking the average value of the fitness of the first scoring index as the index value.

3. The method of scoring a target person of claim 1, further comprising:

acquiring a plurality of second grading indexes;

counting the statistical value of the second scoring index for the target person;

calculating a relative liveness of the second scoring index based on the statistical value of the second scoring index;

and taking the average value of the relative liveness of the second scoring index to obtain the index value.

4. The method of scoring a target person as recited in claim 3, wherein for the second scoring index, the index value is obtained based on:

acquiring the activity of the target character;

calculating an activity multiplier for the target character based on the activity of the target character;

and multiplying the average value of the relative liveness of the second grading index and the liveness multiplier of the target character to obtain the index value.

5. The method of scoring a target person of claim 1, wherein the degree of engagement is cos θ_iThe method comprises the following steps:

${\mathrm{cos\θ}}_i=\frac{{\left(R_i\right)}^T\·R_i^{*}}{\left|R_i\right|\left|R_i^{*}\right|}$

wherein i is the index number, R_iA state distribution matrix for each dimension of the first scoring index,a state distribution matrix (R) for each index of the target person_i)^TIs R_iThe transposed matrix of (2).

6. The method of claim 2, wherein the attention index multiplier is T_iThe method comprises the following steps:

$T_i=\frac{\ln \left(t_i\right)}{\ln \left(\max \right\{t_1,t_2,\mathrm{...},t_n\left\}\right)}$

wherein i is the number of the target person, the attention of the target person, and t₁，t₂，...，t_nAnd n is the number of the other persons and the number of the target persons.

7. The method of claim 3, wherein the relative liveness of the second scoring metric is F_iThe method comprises the following steps:

$F_i=\frac{\ln \left(Y_i\right)}{\ln \left(\max \right\{Y_1,Y_2,\mathrm{...},Y_n\left\}\right)}$

wherein i is the number of the target person, the statistical value of the second score index of the target person, the statistical value of the second score indexes of the other persons and the target person, and n is the number of the other persons and the target person.

8. The method of claim 4, wherein the liveness multiplier is set to S_iThe method comprises the following steps:

$S_i=\frac{\ln \left(g_i\right)}{\ln \left(\max \right\{g_1,g_2,\mathrm{...},g_n\left\}\right)}$

wherein i is the number of the target person, g_iIs the liveness of the target character, g₁，g₂，...，g_nAnd n is the number of the other characters and the target characters.

9. The method of scoring a target person as recited in claim 1, wherein the outputting of the score of the target person based on a multiplication result of all index values by their corresponding index weights comprises:

and accumulating the multiplication results to obtain the score of the target person.

10. The method of scoring a target person as recited in claim 1, wherein the outputting of the score of the target person based on a multiplication result of all index values by their corresponding index weights comprises:

acquiring a third scoring index;

counting the third scoring index value for the target person;

and accumulating the multiplication results and multiplying the accumulated value and the third score index numerical value to obtain the score of the target person.

11. A method of pushing a character, comprising:

scoring all target persons based on the method of any one of claims 1 to 10;

and pushing the target person based on the score of the target person.

12. The method of pushing a character as recited in claim 11, further comprising:

receiving a user character pushing request;

the scoring all target persons is performed based on the request.

13. An apparatus for scoring a target person, comprising:

the acquisition unit is suitable for acquiring a plurality of first scoring indexes and the distribution of dimensionality under the first scoring indexes;

the statistic unit is suitable for counting the state distribution of each dimension of the first grading index aiming at the target person;

the calculating unit is suitable for calculating the engagement degree of the target person on the first scoring index based on the state distribution of each dimension of the first scoring index;

the averaging unit is suitable for averaging the fitness of the first grading indexes belonging to the same index type to obtain an index value;

and the output unit is suitable for outputting the score of the target person based on the multiplication result of all the index values and the corresponding index weights.

14. A system for pushing a person, comprising:

the apparatus of claim 13, adapted to output a score for all target persons;

and the pushing unit is suitable for pushing the target person based on the score of the target person.