CN101685519A - Credit evaluation method and credit evaluation system - Google Patents

Abstract

The invention provides a credit evaluation method and a credit evaluation system. Among them, the credit evaluation system is applicable to different business systems, including: a credit investigation unit, used to obtain the original credit data provided by the business system through the interface with the business system; a data analysis unit, used to analyze the The original credit data obtained by the information unit is used to obtain analysis results; the credit service unit is used to display credit information according to the analysis results obtained by the data analysis unit. Both the credit evaluation method and the credit evaluation system of the present invention can be applied to different business systems. In this way, there is no need to develop a credit evaluation system corresponding to each business system, thereby saving development costs.

Description

Credit Evaluation Method and Credit Evaluation System

technical field

The invention relates to e-commerce technology, in particular to credit evaluation technology.

Background technique

Credit, as an important index in the field of electronic commerce, is more and more concerned by users in the field of electronic commerce. For example, before a user conducts business with the other party through the electronic trading system, he generally needs to know the credit of the other party. After confirming that the credit of the other party meets his own requirements, the user will conduct business with the other party.

As users pay more and more attention to credit, how to conduct credit evaluation has become a hot technical issue in the field of e-commerce. At present, a variety of credit evaluation technologies have emerged in the field of e-commerce. However, these credit evaluation technologies are all developed for specific business systems. For example, an online bookstore corresponds to one credit evaluation system, and a commodity trading platform corresponds to another credit evaluation system. However, these credit evaluation systems cannot be used in common with each other. In this way, for each business system that needs a credit evaluation system, technology developers must develop a large number of credit evaluation systems in order to meet the requirements of a business system corresponding to a credit evaluation system. Require.

It can be seen that developing a credit evaluation system corresponding to each business system will lead to high development costs, so the development costs of existing credit evaluation technologies are relatively high.

Contents of the invention

The technical problem to be solved by the present invention is to provide a credit evaluation method and a credit evaluation system to reduce the development cost of credit evaluation technology.

The present invention provides a credit evaluation method, which is applicable to different business systems, including: obtaining the original credit data provided by the business system through the interface with the business system; analyzing the original credit data to obtain the analysis result; according to The analysis result shows the credit situation.

The present invention provides a credit evaluation system, which is applicable to different business systems, including: a credit investigation unit, used to obtain the original credit data provided by the business system through an interface with the business system; a data analysis unit, used to Analyzing the original credit data obtained by the credit reference unit to obtain an analysis result; a credit service unit for displaying credit status according to the analysis result obtained by the data analysis unit.

Both the credit evaluation method and the credit evaluation system of the present invention can be applied to different business systems. Specifically, the original credit data provided by the business system can be obtained through the interface with the business system, and then the original credit data can be analyzed , and finally show the credit status according to the analysis results. It can be seen that the credit evaluation method and credit evaluation system of the present invention are not only applicable to one business system, but can be applied to many different business systems. In this way, there is no need to develop a credit system for each business system. evaluation system, thereby saving development costs.

Description of drawings

Fig. 1 is a flow chart of a credit evaluation method of the present invention;

Fig. 2 is a schematic structural diagram of a credit evaluation system of the present invention;

FIG. 3 is a schematic structural diagram of the data analysis unit in FIG. 2 .

Detailed ways

First, a credit evaluation method of the present invention is described. This credit evaluation method is applicable to different business systems, and the business system mentioned here refers to a business system that requires credit evaluation. As shown in Figure 1, the credit evaluation method includes:

S101: Obtain the original credit data provided by the business system through an interface with the business system;

S102: Analyze the original credit data to obtain an analysis result;

S103: Present the credit situation according to the analysis result.

The executor of steps S101, S102 and S103 may be a device or system independent of the business system, and this executor may obtain the original credit data through the interface with the business system, wherein the original credit data mentioned here may refer to the business The credit data of the evaluation object obtained by the system, for example, the evaluation status (such as rating) of the evaluation object by users who have contacted with the evaluation object. In practical application, the method shown in Fig. 1 can be applied to any business system, or in other words, the execution subject of steps S101, S102 and S103 can obtain the original credit data through the interface with any business system.

In different business systems, the types of original credit data may be different. For example, the original credit data of some business systems only include the three data of third-party certification, product quality certification and management quality certification, while the original credit data of some business systems only include the two data of single transaction score and product quality certification. In this way, before applying the method shown in FIG. 1 to a business system, it is necessary to specify the type of the obtained original credit data. Of course, the specified type is the type corresponding to the business system. For example, if the method shown in Figure 1 is applied to a business system whose original credit data only include single transaction score and product quality certification, the specified type of original credit data obtained is single transaction score and product quality certified.

In addition, the obtained original credit data can be original credit data for one evaluation object, or original credit data for multiple evaluation objects. For example, the original credit data obtained at a certain time may be the original credit data for enterprise A, or the original credit data for enterprise A and the original credit data for buyer C.

In addition, the obtained original credit data may include static original credit data and dynamic original credit data.

Static original credit data is generally relatively fixed, such as: the score of the completeness of the basic information of the enterprise, the score of the establishment years, the score of the quality of executives, the score of bank credit, the score of certificate certification, etc. Static original credit data can be the value of multiple data after calculation, such as the score of the completeness of the basic information of the enterprise, the score of the establishment years, the score of the quality of executives, the score of bank credit, the score of certificate certification, etc. and.

Dynamic original credit data generally change dynamically. For example, the score for the quality of the service provided by the enterprise changes dynamically along with the quality of the service provided by the enterprise.

The calculation of dynamic credit data generally requires attention to two issues: one is how to score a single transaction; the other is how to aggregate the scores of multiple transactions into dynamic credit data.

For the scoring of a single interaction, the service quality measurement framework can be used. By formulating detailed and reasonable scoring criteria, more accurate scoring results can be obtained. In the embodiment of the present invention, we use a 100-point system for scoring a single interaction, so the highest value of the aggregated dynamic credit data is also 100.

The embodiments of the present invention use correlation, commitment, clarity, and influence (CCCI, Corr _qualities , Commit _criterion , Clear _criterion , Inf _criterion ) methods to score a single interaction. Specifically, each evaluation criterion (evaluation direction) can be customized, such as adding or deleting an evaluation direction, modifying the name of an evaluation direction, and adjusting the weight of each evaluation direction. For each evaluation criterion (evaluation direction), its evaluation scale and description corresponding to each evaluation scale can be configured. For example, for the evaluation direction of "whether the delivery time is timely", it can be configured as -1 point, 0 point, and 1 point for CCCI. At the same time, each score can be configured as "not in time", "general", "Timely" displayed to the user. For the score obtained according to each assessment criterion (assessment direction), it can also be calculated with parameters such as time factor or amount factor, so that the result obtained after the calculation is used as the score corresponding to the assessment criterion (assessment direction). For example, the time factor represents the time of communication, and the amount factor represents the amount of the transaction. The score obtained according to each evaluation criterion (evaluation direction) can be weighted with parameters such as time factor or amount factor, and the weighted result is used as the corresponding Scoring of assessment criteria (assessment direction).

For business systems that can provide static original credit data and dynamic original credit data, the original credit data obtained in step S101 may be the result of calculation between the static original credit data and the dynamic original credit data , for example, assuming that the static original credit data and the dynamic original credit data are respectively a value, then the obtained original credit data may be the value obtained by adding these two values.

In addition, the original credit data obtained in step S101 can be either one-dimensional data or multi-dimensional data. For example, the numerical value obtained by adding the above-mentioned value representing the static original credit data and the value representing the dynamic original credit data is a one-dimensional data, and if the obtained original credit data includes Scoring, then the original credit data obtained is multidimensional data, and the scoring corresponding to an assessment criterion (assessment direction) is one-dimensional data in multidimensional data.

In practical applications, there are many ways to obtain the original credit data provided by the business system. For example, the executor of steps S101, S102 and S103 actively requests the business system to provide original credit data, and the business system provides the original credit data to the executor according to the request. For another example, the business system actively provides original credit data to the execution subject. For another example, the execution subject may obtain the original credit data provided by the business system at regular intervals or multiple times. In this way, in step S102, the execution subject may synthesize the original credit data obtained multiple times for analysis to obtain an analysis result.

In step S102, when analyzing the original credit data, a pre-specified algorithm may be used to analyze the original credit data.

When analyzing the original credit data, it is possible to filter out false original credit data and keep the real original credit data.

There are many ways to filter out the false original credit data and keep the real original credit data. One of the ways is: divide all the original credit data for an evaluation object into two sets, among which, in any set The difference between any one of the original credit data in the same set and other original credit data in the same set is not greater than the difference between any one of the original credit data in the set and the original credit data in another set; according to the preset The rule of filtering out all the original credit data in one of the collections and keeping all the original credit data in the other collection. Wherein, all the original credit data for an evaluation object may include different users' ratings for the evaluation object, or may be multiple ratings for the evaluation object by the same user in multiple interactions. All the original credit data for one evaluation object may be the data obtained once by the execution subject of steps S101, S102 and S103, or the data obtained by the execution subject multiple times. All original credit data for an evaluation object may also refer to all original credit data for the evaluation object in a certain period. In short, all the original credit data for one evaluation object generally refers to multiple original credit data for the evaluation object. In addition, as mentioned above, an original credit data for an evaluation object can be one-dimensional data or multi-dimensional data. In addition, the above preset rules can be determined according to the actual application. For example, it can be stipulated that if all or most of the original credit data in one of the sets are generated within a very close time, then the All or most of the original credit data in this set are false credit data, so when the above two sets are processed according to the pre-set rule, all or most of the original credit data are filtered out It is all the original credit data in the set generated in very close time, and keep all the original credit data in the other set.

The difference between any one original credit data in any one set and other original credit data in the same set is not greater than the difference between any one original credit data in said any one set and the original credit data in another set The implementation method may be: the average value of the difference between any original credit data in any one of the sets and all other original credit data in the same set is not greater than any one of the original credit data in any one of the sets The average of the differences from all raw credit data in another set.

The difference between the above-mentioned original credit data can be the difference between the values of the original credit data, or the difference between the values obtained after the original credit data has been weighted. If the original credit data is a multi-dimensional value The above-mentioned difference between the original credit data can also be the vector distance difference between the original credit data.

The implementation of filtering out false original credit data and retaining real original credit data will be described in detail below through three algorithms respectively. The false original credit data is taken as an example of credit hype data.

1. The first algorithm

First, it is assumed that the normal score conforms to the N(μ, σ) distribution, and the credit hype score conforms to the N(μ′, σ′) distribution, where μ can be regarded as the real credit of the enterprise. In order to achieve the effect of credit hype, it is generally believed that μ′≈100 and σ′≈0. Assume again that in all scoring records, (1-δ) 100% is the real score, and 100δ% is credit hype, then the credit R _nofilter obtained without any processing is:

R _nofilter ≈(1-δ)μ+δμ′

The real credit R _fair is:

R _fair ≈ μ

The difference between R _nofilter and R _fair is called the credit offset value (bias), and the offset value B (μ, σ) is:

B(μ,σ)＝R _nofilter -R _fair ≈δ(100-μ)

This offset value represents the impact of credit speculation on credit. If the offset value is larger, it means that the impact of credit speculation is more serious. The purpose of the first algorithm of the present invention is to reduce B(μ, σ) as much as possible, thereby reducing the influence of credit hype on credit.

Based on the above assumptions, first use the Macnaughton-Smith et al. recursive method to divide the full set N into two clusters, where the full set N includes multiple credit data for one evaluation object:

Step 1: Set two sets A=N, At this stage, elements in A are moved towards B. For each element i in A, first calculate the difference D(i, A={i}) from i to other elements in A:

$\mathrm{<span\; class="notranslate">\; D.</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; \{</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; \}</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; \&NotEqual;</span>\mathrm{<span\; class="notranslate">\; i</span>}}{\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; A</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span>$

Among them, d is a distance function, and the element with the largest D value is moved from A to B first.

Step 2: If there is more than one element in A, then for each remaining element i in A, calculate the difference D(i , A-{i})-D(i, B):

$\mathrm{<span\; class="notranslate">\; D.</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; \{</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; \}</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; D.</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; B</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; \&NotEqual;</span>\mathrm{<span\; class="notranslate">\; i</span>}}{\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; A</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; B</span>}<span\; class="notranslate">\; |</span>}\underset{\mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; B</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; )</span>$

If the result is greater than 0, it means that the difference between i and other elements in A is greater than the difference between i and elements in B, so that i needs to be moved from A to B; if the result is less than 0, it means that i is different from other elements in A The difference between i and the elements in B is less than the difference between i and the elements in B, so that i continues to stay in A; if this result is equal to 0, it means that the difference between i and other elements in A is the same as the difference between i and the elements in B. At this time, you can Keeping i in A can also move i from A to B.

In this way, all scoring results are divided into two mutually disjoint subsets, and the subset with a higher average value is recorded as N _high , and the subset with a lower average value is recorded as N _low . Here, the score in N _high is regarded as the score of credit speculation, and the score in N _low is regarded as the score of real credit. Afterwards, the filtered offset value can be calculated:

B'(μ, σ)=R _{with filter} -R _fair

If the absolute value of B' is less than B, it means that the first algorithm of the present invention has played a positive role in eliminating the influence of credit speculation, so the research on the effect of the first algorithm of the present invention can mainly focus on comparison B' and the size of B.

2. The second algorithm:

其中，n是当前运行算法的时间，m是原始信用数据产生的时间，N是常量，一般是一个比较大的整数。If correction parameters such as time factor or amount factor are used to weight the element i in the first algorithm, and the weighted elements are used to perform operations of several formulas in the first algorithm, it will not be due to the additional time factor Correction parameters such as time factor or amount factor will affect the analysis process. In fact, using correction parameters such as time factor or amount factor will generally only modify the score. Therefore, the second algorithm of the present invention uses correction parameters such as time factor or amount factor to perform weighted operations on element i in the first algorithm, and uses the weighted elements to perform operations on several formulas in the first algorithm. If a time factor is used, the time factor can take

Among them, n is the current running time of the algorithm, m is the time when the original credit data is generated, and N is a constant, generally a relatively large integer.

3. The third algorithm

As mentioned above, the original credit data obtained in step S101 may be multi-dimensional data. For this, when calculating the difference between two multi-dimensional data for the same evaluation object, the difference between the two multi-dimensional data can be calculated. Vector distance. In this way, when some real scoring results and some hyped scoring results are very close, they can be distinguished by the vector distance between the two scoring results.

For example, assume that 5 scoring criteria are taken with the same weight, and each component of the vector represents the score of this interaction under this scoring criterion, and then assume that the scores of the 3 interactions under each scoring criterion are as follows:

[10, 10, 8, 10, 7]

[10, 10, 7, 10, 8]

[8, 8, 10, 10, 10]

Among them, the scoring results of Article 1 and Article 2 are real scores, and the scoring results of Article 3 are credit speculation scores.

Using the vector distance of the scoring result as the difference between the two scoring results, the vector distance between the first scoring result and the third scoring result is:

$\sqrt{{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 10</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 8</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 10</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 8</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 8</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 10</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 10</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 10</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 8</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 10</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 4</span>}$

And the vector distance between the first and second scoring results is:

$\sqrt{{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 10</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 10</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 10</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 10</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 8</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 10</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 10</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 8</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}$

Since the vector distance between the first scoring result and the second scoring result is relatively close, the two real scoring results can be correctly aggregated together.

From the above, we can see that the vector distance function can be expressed as:

${\mathrm{<span\; class="notranslate">\; d</span>}}^{<span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; (</span>\stackrel{<span\; class="notranslate">\; \&Right\; Arrow;</span>}{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; ,</span>\stackrel{<span\; class="notranslate">\; \&Right\; Arrow;</span>}{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span><span\; class="notranslate">\; |</span>\stackrel{<span\; class="notranslate">\; \&Right\; Arrow;</span>}{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; -</span>\stackrel{<span\; class="notranslate">\; \&Right\; Arrow;</span>}{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; |</span>$

和

代表两次不同的交往评分结果的矢量化。in,

and

A vectorization representing the results of two different engagement scores.

Of course, the above three algorithms are only preferred implementation modes given by the present invention. In practical applications, those skilled in the art may also use other feasible algorithms, and such algorithms will not be listed here one by one.

In short, the basic idea of this class of algorithms including the above three algorithms is to divide all scoring results into two sets through a clustering mechanism. Under normal circumstances (assuming that malicious evaluations can be excluded), the set with a higher average score can be regarded as the result of credit speculation, so when analyzing the original credit data, only the set with low scores can be kept. Here, this type of algorithm including the above three algorithms is called clustering and filtering algorithm.

In step S103, there may be multiple ways to present the credit situation, for example, presenting the credit situation in the form of an overall credit profile, a credit report, or a credit trend. Of course, the credit situation can be presented according to a pre-specified presentation manner. In addition, when presenting the credit status according to the analysis result, the analysis result can be directly displayed to the user, or the analysis result can be corresponding to a final credit status, and then the final credit status can be displayed to the user. For example, the analysis result is a score, and the final credit status is a grade. Suppose the analysis result is between 0 and 20 points, then the corresponding grade is "poor", between 21 and 60 is "medium", and between 61 and 80 The average grade is "good", while the grade above 81 points is "excellent".

In addition, after step S102 is performed, the analysis results can also be shared, for example, other business systems can obtain the shared analysis results.

In addition to the above credit evaluation method, the present invention also provides a credit evaluation system, which can be applied to different business systems. As shown in Figure 2, it includes: a credit investigation unit 201, which is used to obtain the original credit data provided by the business system through the interface with the business system; a data analysis unit 202, which is used to analyze the credit information obtained by the credit investigation unit 201. The original credit data of the original credit data, to obtain the analysis result; the credit service unit 203 is used to present the credit situation according to the analysis result obtained by the data analysis unit 202.

Specifically, the credit reference unit 201 can obtain the original credit data through an interface with any service system.

For the introduction of the original credit data obtained by the credit investigation unit 201, reference may be made to the description of the original credit data in the above-mentioned credit evaluation method, which will not be repeated here.

In practical applications, the credit investigation unit 201 may obtain the original credit data provided by the service system in various ways. For example, the credit reference unit 201 actively requests the service system to provide original credit data, and the service system provides the original credit data to the credit reference unit 201 according to the request. For another example, the business system actively provides original credit data to the credit investigation unit 201 . For another example, the credit reference unit 201 may obtain the original credit data provided by the business system at regular intervals or multiple times. In this way, the data analysis unit 202 may analyze the original credit data obtained by the credit reference unit 201 multiple times to obtain the analysis result.

The credit evaluation system may also include a configuration unit 204, which is used to specify that the type of original credit data obtained by the credit investigation unit 201 is related to that of the business system before the credit investigation unit 201 obtains the original credit data provided by the business system. corresponding type. The configuration unit 204 can also be used to specify the algorithm used by the data analysis unit 202 when analyzing the original credit data before the data analysis unit 202 analyzes the original credit data. The configuration unit 204 may also be configured to specify the presentation mode used when the credit service unit 203 presents the credit situation before the credit service unit 203 presents the credit situation.

When analyzing the original credit data, the data analysis unit 202 can filter out false original credit data and keep the real original credit data.

The data analysis unit 202, as shown in FIG. 3 , may include: a data classification unit 2021, configured to divide all original credit data for an evaluation object into two sets, wherein any original credit data in any set and The difference between other original credit data in the same set is not greater than the difference between any original credit data in any one set and the original credit data in another set; the data processing unit 2022 is configured to According to the rules, all the original credit data in one of the sets divided by the data classification unit 2021 are filtered out, and all the original credit data in the other set are retained. Wherein, all the original credit data for an evaluation object may include different users' ratings for the evaluation object, or may be multiple ratings for the evaluation object by the same user in multiple interactions. All the original credit data for an evaluation object may be the data obtained by the credit investigation unit 201 once, or the data obtained by the credit investigation unit 201 multiple times. All original credit data for an evaluation object may also refer to all original credit data for the evaluation object in a certain period. In short, all the original credit data for one evaluation object generally refers to multiple original credit data for the evaluation object. In addition, as mentioned above, an original credit data for an evaluation object can be one-dimensional data or multi-dimensional data. In addition, the above preset rules can be determined according to the actual application. For example, it can be stipulated that if all or most of the original credit data in one of the sets are generated within a very close time, then the All or most of the original credit data in this set are false credit data, so when the data analysis unit 202 processes the above two sets according to the preset rule, all or most of the original credit data will be filtered out. The original credit data are all the original credit data in the set generated in a very close time, and all the original credit data in the other set are kept.

The difference between any one original credit data in any one set and other original credit data in the same set is not greater than the difference between any one original credit data in said any one set and the original credit data in another set The implementation method may be: the average value of the difference between any original credit data in any one of the sets and all other original credit data in the same set is not greater than any one of the original credit data in any one of the sets The average of the differences from all raw credit data in another set.

The difference between the above-mentioned original credit data can be the difference between the values of the original credit data, or the difference between the values obtained after the original credit data has been weighted. If the original credit data is a multi-dimensional value The above-mentioned difference between the original credit data can also be the vector distance difference between the original credit data.

The data analysis unit 202 can use the three algorithms mentioned in the above credit evaluation method to filter out the false original credit data and keep the real original credit data. For specific descriptions, please refer to the description of the three algorithms in the above credit evaluation method, here No longer.

The credit service unit 203 can present the credit situation in various ways, for example, presenting the credit situation in the form of an overall credit profile, a credit report, or a credit trend. Of course, the credit service unit 203 can display the credit information according to the display mode specified by the configuration unit 204 in advance. In addition, when the credit service unit 203 presents the credit status based on the analysis result, it can directly present the analysis result to the user, or can present the final credit status to the user after the analysis result corresponds to a final credit status.

In addition, the credit service unit 203 can also share the analysis results, for example, other business systems can obtain the shared analysis results from the credit service unit 203 .

The credit evaluation system may also include a database 205, which is used to save the original credit data provided by the business system obtained by the credit checking unit 201, provide the original credit data provided by the business system to the data analysis unit 202, and store the original credit data provided by the business system. The data analysis unit 202 obtains the analysis result, and provides the analysis result to the credit service unit 203 .

To sum up, the credit evaluation method and credit evaluation system of the present invention can be applied to different business systems. In the prior art, credit evaluation algorithms are all solidified in the code of the business system. In practical applications, different industries generally require different credit investigation strategies and credit algorithms. For example, the business system of the steel industry and the business system of the food industry need to adopt different credit investigation strategies. Algorithms are applied to the business systems of these two industries, and the program codes of these two business systems must be modified. Obviously, this will inevitably result in a waste of time, manpower and other resources. However, if the credit evaluation method and credit evaluation system of the present invention need to be applied to different industries, they can be put into use only after simple configuration according to the characteristics of the industry, which greatly saves resources.

The above is only a preferred embodiment of the present invention, it should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, some improvements and modifications can also be made, and these improvements and modifications should also be It is regarded as the protection scope of the present invention.

Claims (15)

1. A credit evaluation method is characterized in that the method is suitable for different service systems and comprises the following steps:

acquiring original credit data provided by a service system through an interface between the service system and the service system;

analyzing the original credit data to obtain an analysis result;

and displaying the credit condition according to the analysis result.

2. The credit evaluation method of claim 1, wherein the obtained raw credit data is raw credit data for one evaluation object, or raw credit data for a plurality of evaluation objects, or raw credit data of a type previously designated to correspond to the business system.

3. The credit evaluation method of claim 1, wherein analyzing the raw credit data is performed by: the raw credit data is analyzed using a pre-specified algorithm.

4. The credit evaluation method of claim 1 or 3, wherein the process of analyzing the raw credit data specifically comprises: false original credit data is filtered out, and real original credit data is reserved.

5. The credit evaluation method of claim 4, wherein false raw credit data is filtered out and real raw credit data is retained as follows:

dividing all original credit data aiming at an evaluation object into two sets, wherein the difference between any one original credit data in any one set and other original credit data in the same set is not larger than the difference between any one original credit data in any one set and the original credit data in the other set;

according to a preset rule, all the original credit data in one set are filtered out, and all the original credit data in the other set are reserved.

6. The credit evaluation method of claim 5, wherein the difference between any one original credit data in any one set and other original credit data in the same set is not greater than the difference between any one original credit data in any one set and original credit data in another set by: the average value of the difference between any one original credit data in any one set and all other original credit data in the same set is not larger than the average value of the difference between any one original credit data in any one set and all original credit data in another set.

7. The credit evaluation method of claim 5 or 6, wherein the difference between the original credit data is a difference between values of the original credit data, or a difference between values obtained by weighting the values of the original credit data, or if the original credit data is a data composed of multidimensional values, the difference between the original credit data is a vector distance difference between the original credit data.

8. The credit evaluation method of claim 1, wherein presenting the credit condition is implemented by: and displaying the credit condition according to a pre-specified display mode.

9. A credit evaluation system adapted for use with different business systems, comprising:

the credit investigation unit is used for acquiring original credit data provided by a service system through an interface between the credit investigation unit and the service system;

the data analysis unit is used for analyzing the original credit data obtained by the credit investigation unit to obtain an analysis result;

and the credit service unit is used for showing the credit condition according to the analysis result obtained by the data analysis unit.

10. The credit evaluation system of claim 9, further comprising: the configuration unit is used for appointing the type of the original credit data obtained by the credit investigation unit to be the type corresponding to the service system before the credit investigation unit obtains the original credit data provided by the service system.

11. The credit evaluation system of claim 9, further comprising: a configuration unit for specifying an algorithm used by the data analysis unit when analyzing the raw credit data before the data analysis unit analyzes the raw credit data.

12. The credit evaluation system of claim 9, wherein the data analysis unit filters out false raw credit data and retains true raw credit data when analyzing the raw credit data.

13. The credit evaluation system of claim 12, wherein the data analysis unit comprises:

the data classification unit is used for dividing all original credit data aiming at an evaluation object into two sets, wherein the difference between any one original credit data in any one set and other original credit data in the same set is not larger than the difference between any one original credit data in any one set and the original credit data in the other set;

and the data processing unit is used for filtering out all original credit data in one set divided by the data classification unit according to a preset rule and reserving all original credit data in the other set.

14. The credit evaluation system of claim 9, further comprising: the configuration unit is used for appointing the showing mode used when the credit service unit shows the credit condition before the credit service unit shows the credit condition.

15. The credit evaluation system of claim 9, further comprising: and the database is used for storing the original credit data provided by the service system and obtained by the credit investigation unit, providing the original credit data provided by the service system to the data analysis unit, storing the analysis result obtained by the data analysis unit and providing the analysis result to the credit service unit.

Images