CN110472025B

CN110472025B - Method, device, computer equipment and storage medium for processing session information

Info

Publication number: CN110472025B
Application number: CN201910635361.6A
Authority: CN
Inventors: 吴冶成; 蒋逸文; 叶曙峰; 黄鸿顺
Original assignee: Ping An Technology Shenzhen Co Ltd
Current assignee: Ping An Technology Shenzhen Co Ltd
Priority date: 2019-07-15
Filing date: 2019-07-15
Publication date: 2024-01-30
Anticipated expiration: 2039-07-15
Also published as: CN110472025A

Abstract

The application relates to a session message processing method, a session message processing device, computer equipment and a storage medium. According to the method, when the session information in the session interface comprises target text information, a product identification code corresponding to the target text information and a product to-be-predicted index are obtained, so that numerical resource data corresponding to the product identification code are obtained from a database, the numerical resource data are converted into associated values corresponding to the product to-be-predicted index according to a data conversion instruction and are input into a linear regression model, a numerical attribute predicted value of the product to-be-predicted index is obtained, further response information corresponding to the session information is generated according to the numerical attribute predicted value, the numerical attribute predicted value corresponding to the product identification is calculated quickly, an important basis is provided for a user to select a product, flexible information inquiry is realized, and the pertinence of further information inquiry is poor and the accuracy is low.

Description

Method, device, computer equipment and storage medium for processing session information

Technical Field

The present invention relates to the field of data processing technologies, and in particular, to a method and apparatus for processing session information, a computer device, and a storage medium.

Background

With the rapid development of internet technology, intelligent question-answering systems have been popularized to various industries. At present, the dialogue of the intelligent question-answering system is designed based on standard question-answering pairs, generally the questions input by users, and the intelligent question-answering system queries standard contents matched with the questions input by the users from a pre-established database and sends the standard contents obtained by matching to a user side for answering. The traditional intelligent question-answering system can only answer in a template mode, and has poor pertinence and low accuracy for different information inquiry.

Disclosure of Invention

In view of the foregoing, it is desirable to provide a method, apparatus, computer device, and storage medium for processing session information.

A method of processing session information, the method comprising:

acquiring a session message in a session interface, and acquiring a product identification code and a product to-be-predicted index corresponding to target text information when the session message comprises the target text information;

acquiring numerical resource data corresponding to the product identification code from a database;

acquiring a data conversion instruction, and converting the numerical resource data into an associated value corresponding to the product to-be-predicted index according to the data conversion instruction;

Inputting the associated value corresponding to the product to-be-predicted index into a linear regression model to obtain a numerical attribute predicted value of the product to-be-predicted index;

and generating response information corresponding to the session message according to the numerical attribute predicted value and a preset format, and displaying the response information through the session interface.

In one implementation, the step of converting the numerical resource data into the associated value corresponding to the index to be predicted of the product according to the data conversion instruction includes:

extracting a data conversion rule in the preset data conversion instruction; the data conversion rule is a conversion rule between the numerical resource data and the associated value;

acquiring data attributes in the data conversion rule, and acquiring target data corresponding to the data attributes from the numerical resource data;

and converting the obtained target data corresponding to the data attribute according to the data conversion rule to obtain the association value corresponding to the product to-be-predicted index.

In one implementation, the step of inputting the association value corresponding to the product to-be-predicted index into a linear regression model to obtain the numerical attribute predicted value of the product to-be-predicted index includes:

Acquiring a time stamp of the session message, and determining a target time period according to the time stamp and the reference window length of the linear regression model;

acquiring a numerical attribute value corresponding to the product to-be-predicted index in a target time period, and inputting an association value corresponding to the product to-be-predicted index in the target time period and the numerical attribute value into a linear regression model to train the linear regression model;

and inputting the correlation value of the time stamp corresponding to the time into a trained linear regression model, and outputting the numerical attribute predicted value of the product to be predicted index by the trained linear regression model.

In one implementation, before the step of obtaining the timestamp of the session message, the method further includes:

acquiring a numerical attribute value and an associated value corresponding to the product to-be-predicted index in a first time period;

acquiring a plurality of different pre-selected regression window lengths, and determining a second time period according to each pre-selected regression window length;

respectively constructing a preselected linear regression model corresponding to the length of each preselected regression window according to the numerical attribute value and the associated value corresponding to the product to be predicted index in each second time period;

Respectively inputting the association values of the first time period into a preselected linear regression model corresponding to each preselected window length to obtain numerical attribute predicted values of a plurality of product to-be-predicted indexes in the first time period;

and respectively calculating error values between the numerical attribute predicted values and the numerical attribute values obtained through the pre-selected linear regression models corresponding to the pre-selected window lengths, and taking the pre-selected regression window length corresponding to the numerical attribute predicted value with the smallest error of the numerical attribute values as a reference regression window length.

In one implementation, the step of using the pre-selected regression window length corresponding to the numerical attribute predicted value with the smallest error of the numerical attribute value as the reference regression window length includes:

acquiring an adjustment coefficient of the length of the reference window;

adjusting the length of the reference regression window according to the adjustment coefficient to obtain the optimal regression window length;

and determining the optimal regression window length as a reference window length.

In one implementation, the step of obtaining the product identifier corresponding to the target text information and the product to-be-predicted index includes:

matching the target text information with a known product identifier;

And if the target text information is matched with the known product identifier, determining the product identifier of the known product identifier as the product identifier corresponding to the target text information.

A processing apparatus of session information, the apparatus comprising:

the session information acquisition module is used for acquiring a session message in a session interface, and acquiring a product identification code and a product to-be-predicted index corresponding to target text information when the session message comprises the target text information;

the numerical resource data acquisition module is used for acquiring numerical resource data corresponding to the product identification code from a database;

the association value acquisition module is used for acquiring a data conversion instruction, and converting the numerical resource data into association values corresponding to the product to-be-predicted indexes according to the data conversion instruction;

the predicted value acquisition module is used for inputting the associated value corresponding to the product to-be-predicted index into a linear regression model to obtain a numerical attribute predicted value of the product to-be-predicted index;

and the response information generation module is used for generating response information corresponding to the session message according to the numerical attribute predicted value and a preset format, and displaying the response information through the session interface.

In one implementation, the association value obtaining module is specifically configured to: extracting a data conversion rule in the preset data conversion instruction; the data conversion rule is a conversion rule between the numerical resource data and the associated value; acquiring data attributes in the data conversion rule, and acquiring target data corresponding to the data attributes from the numerical resource data; and converting the obtained target data corresponding to the data attribute according to the data conversion rule to obtain the association value corresponding to the product to-be-predicted index.

A computer device comprising a memory storing a computer program and a processor which when executing the computer program performs the steps of:

A computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of:

According to the method, the device, the computer equipment and the storage medium for processing the session information, the session information in the session interface is obtained, when the session information comprises the target text information, the product identification code corresponding to the target text information and the product to-be-predicted index are obtained, so that the numerical resource data corresponding to the product identification code are obtained from the database, the numerical resource data are converted into the associated value corresponding to the product to-be-predicted index according to the data conversion instruction and are input into the linear regression model, the numerical attribute predicted value of the product to-be-predicted index is obtained, further the response information corresponding to the session information is generated according to the numerical attribute predicted value, the important basis is provided for a user to select the product, the flexible information query is realized, and the pertinence of the further information query is poor and the accuracy is low.

Drawings

FIG. 1 is an application scenario diagram of a method for processing session information in one embodiment;

FIG. 2 is a flow chart of a method for processing session information in one embodiment;

FIG. 3 is a flow chart of inputting a correlation value corresponding to a product to-be-predicted index into a linear regression model to obtain a numerical attribute predicted value of the product to-be-predicted index in one embodiment;

FIG. 4 is a block diagram showing a structure of a processing apparatus for session information in one embodiment;

fig. 5 is an internal structural diagram of a computer device in one embodiment.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, the present application will be further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the present application.

The session information processing method provided by the application can be applied to an application environment shown in fig. 1. Wherein the terminal 102 communicates with the server 104 via a network. The terminal 102 is provided with an intelligent question-answering application program, a user can realize product information data query in a session form through the intelligent question-answering application program, specifically, the terminal 102 can display a session interface for a session with the intelligent question-answering system, the user inputs a problem through the session interface displayed by the terminal 102, the problem input by the user is sent to the server 104 as a session message, when the session message acquired by the server 104 comprises target text information, the server 104 determines a product identification code corresponding to the target text information and a product to-be-predicted index according to the target text information, numerical resource data corresponding to the product identification code is acquired from a database, after the numerical resource data is converted into an associated value corresponding to the product to-be-predicted index according to a data conversion instruction, the associated value corresponding to the product to-be-predicted index is input into a linear regression model, the numerical attribute predicted value of the product to be-predicted index is obtained, accordingly, the response information corresponding to the session message is generated according to a preset format by the numerical attribute predicted value, and the session interface response information displayed by the terminal 102 is displayed. The terminal 102 may be, but not limited to, various personal computers, notebook computers, smartphones, tablet computers, and portable wearable devices, and the server 104 may be implemented by a stand-alone server or a server cluster composed of a plurality of servers.

In one embodiment, as shown in fig. 2, a method for processing session information is provided, and the method is applied to the server in fig. 1 for illustration, and includes the following steps:

step S210: and acquiring a session message in the session interface, and acquiring a product identification code and a product to-be-predicted index corresponding to the target text information when the session message comprises the target text information.

In this step, the session refers to a process that a certain user performs information interaction with other users, and the session page is an interface for displaying session messages, where the session messages are information input by the user and collected during the session, and the session messages may include text information, voice information and the like sent by the session object.

The target text information refers to preset texts, such as enterprise names, project names and the like; the product identification code is used for uniquely identifying a character string of the product to be queried; the product to-be-predicted index is used for indicating a guiding value of influence of different data resources of the item to be queried on the investment risk of the item; the terminal establishes a mapping relation table between the target text information and different products in advance, and when the server detects the target text information from the session message, the terminal can determine the product identification code corresponding to the target text information according to the mapping relation between the target text information and the different products, and acquire the index to be predicted of the product identification code. For example, the target text information includes an enterprise name of the a enterprise, and when the terminal detects the enterprise name of the a enterprise in the session message, all product identification codes of the a enterprise can be obtained according to mapping of the enterprise name of the a enterprise, and a corresponding index to be predicted is determined.

In a specific implementation, the terminal is provided with an intelligent question-answering application program, and a user can inquire product information such as numerical attribute data of bond products such as bond products and bond products through the intelligent question-answering application program. For example, the user inputs a question consulted with the intelligent question-answering system based on a session message input component in a session interface of the intelligent question-answering application, and clicks a send button; the terminal generates a session message according to the problem input by the user, sends the generated session message to the server, and receives the session message, and when the session message is identified to include target text information, obtains a product identification code and a product to-be-predicted index corresponding to the target text information so as to determine a target product and target product information inquired by the user.

In one embodiment, the step of obtaining the product identification code and the product to-be-predicted index corresponding to the target text information includes: matching the target text information with the known product identifier; if the target text information is matched with the known product identifier, determining the product identifier of the known product identifier as the product identifier corresponding to the target text information.

Step S220: and acquiring numerical resource data corresponding to the product identification code from a database.

The numerical resource data refers to data affecting the index to be predicted of the product, for example, taking the product corresponding to the product identification code as a bond product, and the numerical resource data may be different index data of each day of the bond product, for example, numerical resource data such as the mcholy duration, the numerical gain rate of the current day, and the like of the bond market index.

Specifically, the terminal uses the name of the product corresponding to the product identification code as a search keyword based on big data, crawls the corresponding data from the internet in advance, performs preprocessing operation on the crawled data, acquires the data content of the numerical value type, and stores the data content as the numerical value resource data corresponding to the product identification code, so that a subsequent server can conveniently acquire the corresponding numerical value resource data by querying a database.

Step S230: and acquiring a data conversion instruction, and converting the numerical resource data into an associated value corresponding to the index to be predicted of the product according to the data conversion instruction.

In the step, the data conversion instruction is an instruction capable of converting numerical resource data into an associated value corresponding to a product to-be-predicted index; the associated value refers to a value corresponding to the index to be predicted of the product, which is converted from the numerical resource data based on the data conversion instruction. For example, the numerical resource data such as the mcholy period and the numerical gain rate of the bond market index are converted into a interest factor value reflecting the structural factor of the interest period of the bond product.

In the specific implementation, the server acquires a preset data conversion instruction, converts the acquired numerical resource data according to the preset data conversion instruction to acquire different values, and recognizes the acquired level to be predicted as a correlation value corresponding to the resource data prediction index, so that the subsequent server can conveniently predict the resource data prediction value according to the acquired correlation value, and the data volume is effectively reduced.

Step S240: and inputting the associated value corresponding to the product to-be-predicted index into a linear regression model to obtain the numerical attribute predicted value of the product to-be-predicted index.

In the step, a linear regression model is trained by utilizing a historical numerical attribute value and a historical association value of a product to be predicted index in advance; after obtaining the associated values of the product to be predicted indexes of the product to be queried, the server inputs the associated values corresponding to the product to be predicted indexes of the product to be queried into a linear regression model, and the linear regression model outputs the numerical attribute predicted values of the product to be predicted indexes, so that the accuracy of the obtained resource data predicted information can be effectively improved.

Step S250: and generating response information corresponding to the session message according to the numerical attribute predicted value in a preset format, and displaying the response information through a session interface.

In the step, the server acquires the preset format of the response information, fills the numerical attribute predicted value corresponding to the product identifier into the preset format, generates the response information corresponding to the session information, returns the generated response information to the terminal, displays the response information through the session interface of the terminal, facilitates the user to obtain the numerical attribute predicted value of the queried product, and is beneficial to the user to invest in the system and the like according to the numerical attribute predicted value.

According to the session information processing method, the session information in the session interface is obtained, when the session information comprises the target text information, the product identification code corresponding to the target text information and the product to-be-predicted index are obtained, so that the numerical resource data corresponding to the product identification code are obtained from the database, the numerical resource data are converted into the associated value corresponding to the product to-be-predicted index according to the data conversion instruction and are input into the linear regression model, the numerical attribute predicted value of the product to-be-predicted index is obtained, further the session information corresponding response information is generated according to the numerical attribute predicted value, the numerical attribute predicted value corresponding to the product identification is calculated rapidly, an important basis is provided for a user to select a product, flexible information inquiry is realized, and the pertinence of further information inquiry is poor and the accuracy is low.

In one embodiment, the step of converting the numerical resource data into the associated value corresponding to the index to be predicted of the product according to the data conversion instruction includes: extracting a data conversion rule in a preset data conversion instruction; the data conversion rule is a conversion rule between the numerical resource data and the associated value; acquiring data attributes in a data conversion rule, and acquiring target data corresponding to the data attributes from the numerical resource data; and converting the obtained target data corresponding to the data attribute according to the data conversion rule to obtain the association value corresponding to the product to-be-predicted index.

In this embodiment, the data types of the association values corresponding to the product to be predicted index are multiple, and the conversion rules of the association values of different data types are different; the server acquires a plurality of conversion rules from the data conversion instruction, analyzes the conversion rules, acquires the data attribute of the data used in each conversion rule, acquires the corresponding target data from the numerical resource data according to the data attribute, and substitutes the acquired target data with different data attributes into the corresponding conversion rules to acquire the association value corresponding to the index to be predicted of the converted product.

Taking a product corresponding to the product identification code as a bond product and a product to-be-predicted index as a bond net value of the bond product as an example, wherein the data attribute in the data conversion rule comprises a long-term mcdona of the bond market index and a daily gain rate of the bond market index, and after acquiring numerical value data corresponding to the long-term mcdona of the bond market index and the daily gain rate of the bond market index, the server determines the configuration weight of each bond market index according to the numerical value data of the long-term mcdona of each bond market index, and further calculates the correlation value of the first type according to the configuration weight and the numerical value data of the daily gain rate.

Wherein the first type of associated values may include slope factor values, convexity factor values, credit risk factor values, violation factor values, and the like; for example, the server determines a slope factor value from a daily gain ratio of a long-term neutral and a convex neutral combination of the medium-liability long-term bond total financial index and the medium-liability short-term bond total financial index, determines a convex factor value from a daily gain ratio of a long-term neutral and a convex neutral combination of five indexes of the medium-liability total financial index (1 to 3 years, 3 to 5 years, 5 to 7 years, 7 to 10 years, and more than 10 years), determines a credit risk factor value from a daily gain ratio of a long-term neutral combination of the medium-liability AAA grade financial index and the medium-liability national opening total financial index, and determines a violation factor value from a daily gain ratio of a long-term neutral combination of the medium-liability high-liability index and the medium-liability AAA grade financial index.

Taking the calculated slope factor value as an example, the server extracts the long term d of the total financial index of the long-term bond of the medium-term bond from the data to be predicted ₁ And a long period d of the medium-short term bond total financial index of the medium-short term bond ₂ Let w is ₁ And w ₂ The configuration weights of the long-term bond total financial index of the medium debt and the short-term bond total financial index of the medium debt are respectively, and then the sum of the long-term bond total financial index of the medium debt and the short-term bond total financial index of the medium debt is according to w ₁ +w ₂ ＝1，w ₁ *d ₁ +w ₂ *d ₂ And (0) obtaining the configuration weight of the long-term combination of the two bond market indexes, so as to calculate the total financial index of the long-term bonds of the medium-term bonds and the daily gain rate of the long-term combination of the total financial index of the short-term bonds of the medium-term bonds according to the configuration weight and the daily gain rate.

Likewise, the data attributes in the data transformation rules further include a numerical gain rate of the bond market index and a target rate, wherein the target rate refers to a risk-free interest rate on the bond market and can be selected as a short-term national liability rate. The server may calculate a second type of associated value based on the numerical gain rate and the risk-free interest rate of the bond market index, wherein the second type of associated value may include a interest factor value, a monetary factor value, and a rotatable market factor value, e.g., the interest factor value is determined based on a difference between a medium-debt total financial-rich index daily gain rate and the risk-free interest rate, the monetary factor value is determined based on a difference between a medium-certificate monetary-bond index daily gain rate and the risk-free interest rate, and the rotatable market factor value is determined based on a difference between a medium-certificate rotatable debt index and the risk-free interest rate.

In the above embodiment, based on the basic principle of the bond product, from the perspective of profit decomposition of the bond product, the profit of the bond product is decomposed into a plurality of different profit influencing factors as the associated values of the product to be predicted index of the bond net value, and the data conversion rule of each associated value is set through the policy of real disc operation; and calculating the association value of the product to-be-predicted index by utilizing the data conversion rule, and realizing nonlinear combination optimization processing on the market index, so that the prediction of the bond net value of the bond product is closer to the common operation and characteristics of the bond product, and the accuracy of estimating the product to-be-predicted index is improved.

In one embodiment, as shown in fig. 3, fig. 3 is a flow chart of inputting a correlation value corresponding to a product to-be-predicted index into a linear regression model to obtain a numerical attribute predicted value of the product to-be-predicted index, and inputting a correlation value corresponding to the product to-be-predicted index into the linear regression model to obtain a numerical attribute predicted value of the product to-be-predicted index, where the method includes:

Step S310: acquiring a time stamp of the session message, and determining a target time period according to the time stamp and the reference window length of the linear regression model;

step S320: and acquiring a numerical attribute value corresponding to the product to-be-predicted index in the target time period, and inputting the association value and the numerical attribute value corresponding to the product to-be-predicted index in the target time period into the linear regression model to train the linear regression model.

Step S330: and inputting the correlation value of the time stamp corresponding to the time into a trained linear regression model, and outputting the numerical attribute predicted value of the index to be predicted of the product by the trained linear regression model.

In this embodiment, the linear regression model is expressed asWherein y is a numerical attribute predicted value of the product to be predicted index, x is each associated value of the to-be-predicted index, and θ is a corresponding coefficient of each associated value; the time stamp refers to the date of the processing of the session information; the reference regression window length refers to the value length of the correlation value of the linear regression model, and the target time period is a historical time period which takes the receiving time as the starting time and is forward pushed and takes the duration as the standard window length; specifically, the server obtains the product within the target time period After the correlation value and the numerical attribute value corresponding to the to-be-predicted index of the product are obtained, training the coefficient of the linear regression model by using the correlation value and the numerical attribute value corresponding to the to-be-predicted index of the product in the target time period, and after training, inputting the correlation value of the time stamp corresponding to the time into the linear regression model to obtain the value output by the linear regression model as the numerical attribute predicted value of the to-be-predicted index of the product.

Still taking the product corresponding to the product identification code as a bond product and the bond net value of the bond product as a product to be predicted index, the server determines the historical time period with the time length as the reference regression window length as a target time period according to the time stamp of the session message and the reference regression window length, for example, the reference regression window length can be set to 35 days, when the numerical value attribute of a certain bond 9 months and 29 days is required to be estimated, the target time period is determined to be thirty-five transaction days from 8 months and 25 days to 9 months and 28 days, and when the numerical value attribute of the bond at 9 months and 28 days is required to be estimated, the target time period is determined to be thirty-five transaction days from 8 months and 24 days to 9 months and 27 days. After obtaining the correlation value corresponding to the bond net value and the history unit net value corresponding to the bond net value, the server substitutes the history correlation value and the history unit net value of each day in the target time period into the linear regression model one by one as training data, minimizes the mean square error in the target time period as an optimization target, calculates the weight coefficient of each correlation value, obtains the trained linear regression model, obtains each correlation value of the bond product at the time stamp, and calculates the bond net value predicted value of the bond product at the time stamp.

Because the linear regression model predicts the numerical attribute predicted value of the product to be predicted according to the historical correlation value, the regression window length has great influence on the accuracy of the numerical attribute predicted value, the accuracy of the numerical attribute predicted value can be obviously improved by the proper window length, and the accuracy of the numerical attribute predicted value can be greatly reduced by the improper window length. Thus, in one embodiment, before the step of obtaining the timestamp of the session message, further comprises: acquiring a numerical attribute value and an associated value corresponding to a product to-be-predicted index in a first time period; acquiring a plurality of different pre-selected regression window lengths, and determining a second time period according to each pre-selected regression window length; respectively constructing a preselected linear regression model corresponding to the length of each preselected regression window according to the numerical attribute value and the associated value corresponding to the product to be predicted index in each second time period; respectively inputting the correlation values of the first time period into a preselected linear regression model corresponding to each preselected window length to obtain numerical attribute predicted values of a plurality of product to-be-predicted indexes in the first time period; and respectively calculating the error value between the numerical attribute predicted value and the numerical attribute value obtained through the pre-selected linear regression model corresponding to each pre-selected window length, and taking the pre-selected regression window length corresponding to the numerical attribute predicted value with the minimum error of the numerical attribute value as the reference regression window length.

In this embodiment, the first time period may be selected as the first 5 days of the time stamp, for example, the time stamp is 12 months and 20 days, the first time period is 12 months and 15 days to 12 months and 19 days, and the server obtains the numerical resource data of the sample products daily in 12 months and 15 days to 12 months and 19 days as the sample value.

The preselected window length refers to the value length of the correlation value of the linear regression model; after obtaining the first time period and the plurality of preselected window lengths, the server obtains historical time periods with the date of the first time period as the starting time and the duration of different preselected window lengths, and obtains numerical attribute values of product to-be-predicted indexes of sample products on each day in the historical time periods and all associated values.

For example, the server may set the preselected window lengths of 10 days, 20 days, 30 days, …, and 100 days respectively with a step length of 10 days, and after obtaining the first time period of 12 months 15 days to 12 months 19 days, the server determines the second time period of 10 days to be 12 months 5 days to 12 months 14 days, and obtains the numerical attribute value of the product to be predicted index in the second time period and each associated value; similarly, the server obtains the numerical attribute values of the product to be predicted indexes and the associated values in the second time periods with different preselected window lengths, such as 20 days, 30 days and the like, respectively, by determining that the second time period with the 20 days is 11 months, 25 days, 12 months and 14 days, and the like.

When the time window length is 10 days, training a linear regression model through the numerical attribute values of the product to be predicted indexes in the historical time period with the time length of 10 days and the associated values; when the length of the time window is 20 days, training a linear regression model through the numerical attribute values and the associated values of the product to be predicted indexes in the historical time period with the time length of 20 days, and the like, wherein the server obtains preselected linear regression models corresponding to a plurality of preselected window lengths.

The server inputs each associated value of each day in the first time period to a preselected linear regression model respectively to obtain a numerical attribute predicted value of a product to be predicted index corresponding to each day, so that the numerical attribute predicted value of the product to be predicted index predicted by using the preselected linear regression model corresponding to different preselected window lengths in each day can be obtained; and the server determines the length of the reference window according to the error value between the numerical attribute value of the product to be predicted index of each day in the first time period and the numerical attribute predicted value of the product to be predicted index of each day predicted by different preselected linear regression models. In this embodiment, the numerical attribute predicted value error of a single sample product is minimized to be an optimization target in a first time period, and an initial value of the optimal regression window length is selected, so that the accuracy of the numerical attribute predicted value is improved.

In one embodiment, the step of taking the preselected regression window length corresponding to the numerical attribute predictor with the smallest error of the numerical attribute value as the reference regression window length comprises: acquiring an adjustment coefficient of the length of a reference window; adjusting the length of the reference regression window according to the adjustment coefficient to obtain the optimal regression window length; the optimal regression window length is determined as the reference window length.

When the prediction model is a linear regression model, the accuracy of the predicted value of the predicted numerical attribute of the linear regression model is often related to the length of a regression window; the adjustment coefficient of the reference regression window length can be set by a user, and can be sent to the server through the terminal, or can be set after the server analyzes the digital resource data. Specifically, after the server obtains the adjustment coefficient of the reference window, the length of the reference regression window is adjusted according to the adjustment coefficient to obtain the optimal regression window length, and then the optimal regression window length is used as the regression window length to acquire the numerical attribute predicted value of the index to be predicted, so that the acquisition accuracy of the numerical attribute predicted value of the index to be predicted is further improved.

Taking a bond product as an example, when the server obtains a numerical attribute predicted value of an index to be predicted of the product, for example, a predicted value of a bond net value, the adjustment coefficients include a public opinion adjustment coefficient, a scenic spot adjustment coefficient and an economic cycle adjustment coefficient of the bond product.

For the public opinion adjustment coefficient, since the linear regression model predicts the value attribute predicted value of the bond product according to the correlation value affecting the net value of the unit, the public opinion condition affecting the magnitude of the value attribute predicted value generally occurs more suddenly, and the influence on the value attribute predicted value is larger, when the regression window length is too long, the linear regression model is difficult to quickly react to the public opinion condition, resulting in the reduction of the accuracy of net value estimation of the linear regression model. Therefore, the server can acquire the business name to which the bond product belongs and the business name to which the bond product belongs; according to the enterprise name and the industry name, news corpus data of the bond products in a preset time period is obtained; extracting event keywords from news corpus data, and matching the event keywords with event tags in a preset event tag library; and determining a public opinion adjustment coefficient according to the matching result of the event keywords and the event labels.

The preset time period is a period of time before the preset receiving time, and may be one month, half month, etc. The server obtains keywords related to bond products, crawls news corpus data related to the keywords from the internet through a web crawler technology, analyzes the news corpus data by utilizing a TF-IDF (term frequency-inverse document frequency) algorithm, extracts event keywords capable of summarizing the news corpus data from each piece of news corpus data, and determines that major public opinion accidents occur in related enterprises or industries of the bond products to be shared when matching results of the event keywords and the event tags are successful by matching the event keywords with a preset event tag library, wherein at the moment, the server can set a public opinion adjustment coefficient to a value smaller than 1 to realize the reduction of the length of a reference regression window so as to reduce the influence of historical unit net values of target time periods and historical association values affecting the unit net values on numerical attribute predicted values of the bond net values, so that the accuracy of the numerical attribute predicted values is improved; when the matching result of the event keywords and the event labels fails to match, it is determined that no major public opinion accident occurs in the related enterprises or industries of the bond products to be shared, and at this time, the server does not set the public opinion adjustment coefficient or sets the public opinion adjustment coefficient to 1. Setting a public opinion adjustment coefficient according to the matching result of event keywords of the news corpus data and event tags in a preset event tag library by acquiring the news corpus data related to the bond products, and subsequently adjusting the regression window length according to the adjustment coefficient to adjust the regression window length of the linear regression model according to the public opinion condition of the products, so that the accuracy of the model on the numerical attribute predicted value of the product to be predicted index of the single bond product is improved.

For the scenic spot strength adjustment coefficient, the server can acquire the historical financial index of the bond product to be shared from the object source data, and acquire the tertile number of the historical financial index; acquiring financial indexes of the bond products to be shared at the receiving time; comparing the financial index with the tertile number to determine the type of the industrial scene; and determining a scenic spot type adjustment coefficient according to the industrial scenic spot type. Specifically, the server obtains the historical financial index of the bond product to be shared, and obtains a first trisection number Q1 and a second trisection number Q2 of the historical financial index, wherein the first trisection number Q1 and the second trisection number Q2 divide all the historical financial index into trisections respectively, and the three trisections are respectively a high value, a medium value and a low value. After the server obtains the financial index of the receiving time, comparing the financial index of the receiving time with the third decimal place, when the financial index of the receiving time is larger than the first third decimal place Q1 and smaller than the second third decimal place Q2, the industrial view gas type of the receiving time is considered to be a stable type, when the financial index of the receiving time is smaller than the first third decimal place Q1, the current industrial view gas type is considered to be a declining type, and when the financial index of the receiving time is larger than the second third decimal place Q2, the current industrial view gas type is considered to be a view gas type. The financial index can be selected as the rolling market rate of the bond products, and the rolling market rate of the bond products can be used for accurately estimating the scenic spot type of the current industry.

When the current industry scenic spot type is a scenic spot type or a fading type, the server can set the scenic spot type adjustment coefficient to a value smaller than 1, so as to reduce the length of a reference regression window, reduce the influence of the historical unit net value of the target time period and the historical association value affecting the unit net value on the numerical attribute predicted value of the bond net value, and improve the accuracy of the numerical attribute predicted value; when the current industry scenic spot type is a stable type, the server does not set the scenic spot type adjustment coefficient or sets the scenic spot type adjustment coefficient to 1, and the reference regression window length is not adjusted. The current industry scene air degree is determined by acquiring the financial index of the receiving time, when the industry scene air degree is in scenic spots or decays, the length of a reference regression window is shortened, so that the influence of the historical unit net value of a target time period and the historical association value affecting the unit net value on the numerical attribute predicted value of the bond net value is reduced, the accuracy of the numerical attribute predicted value is improved, the regression window length of a linear regression model is adjusted according to the market environment corresponding to the bond product, and the accuracy of the numerical attribute predicted value of the product to be predicted index of the bond product is improved.

For the economic cycle adjustment coefficient, the server acquires a stock market index in a preset time period and generates an index curve according to the stock market index; the index curve is sent to a pre-trained neural network model, and the economic cycle type is obtained through the neural network model; and determining an economic cycle adjustment coefficient according to the economic cycle type.

Wherein, the economic cycle type can comprise mild rising, mild falling, rising, falling and the like; the server pre-builds a neural network model that can be pre-trained with a plurality of exponential curves with economic cycle-type labels. The server obtains a stock market index, such as a lushen 300 index and the like, in a preset time period, generates an index curve according to the stock market index, and obtains the economic cycle type by using the neural network model through adding the index curve into the neural network model. When the current economic cycle adjustment coefficient is of a storm or a storm type, the server can set the economic cycle adjustment coefficient to a value smaller than 1, so as to reduce the length of a reference regression window, reduce the influence of the historical unit net value of the target time period and the historical association value affecting the unit net value on the numerical attribute predicted value of the bond net value, and improve the accuracy of the numerical attribute predicted value; when the current economic cycle is of the gentle rising or gentle falling type, the server does not set the economic cycle adjustment coefficient or sets the economic cycle adjustment coefficient to 1, and the reference regression window length is not adjusted. By acquiring the type of the economic cycle in which the receiving time is located, when the economic cycle is in a fluctuation cycle, the length of the reference regression window is shortened, the regression window length of the linear regression model is adjusted according to the current macroscopic economic environment, and the accuracy of the model on the numerical attribute predicted value of the product to be predicted index of the single bond product is improved.

It should be understood that, although the steps in the flowcharts of fig. 2 to 4 are sequentially shown as indicated by arrows, these steps are not necessarily sequentially performed in the order indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least some of the steps in fig. 2-4 may include multiple sub-steps or stages that are not necessarily performed at the same time, but may be performed at different times, nor does the order in which the sub-steps or stages are performed necessarily occur in sequence, but may be performed alternately or alternately with at least a portion of other steps or sub-steps or stages of other steps.

In one embodiment, as shown in fig. 4, there is provided a processing apparatus of session information, including: the system comprises a session message acquisition module, a numerical resource data acquisition module, an association value acquisition module, a predicted value acquisition module and a response information generation module, wherein:

a session message obtaining module 410, configured to obtain a session message in a session interface, and when the session message includes target text information, obtain a product identifier corresponding to the target text information and a product to-be-predicted index;

The numerical resource data obtaining module 420 is configured to obtain numerical resource data corresponding to the product identifier from the database;

the association value obtaining module 430 is configured to obtain a data conversion instruction, and convert the numerical resource data into an association value corresponding to the index to be predicted of the product according to the data conversion instruction;

the predicted value obtaining module 440 is configured to input an associated value corresponding to the product to-be-predicted index into the linear regression model, so as to obtain a numerical attribute predicted value of the product to-be-predicted index;

the response information generating module 450 is configured to generate response information corresponding to the session message according to the numerical attribute predicted value and a preset format, and display the response information through the session interface.

In one embodiment, the association value obtaining module is specifically configured to: extracting a data conversion rule in a preset data conversion instruction; the data conversion rule is a conversion rule between the numerical resource data and the associated value; acquiring data attributes in the data conversion rule, and acquiring target data corresponding to the data attributes from the numerical resource data; and converting the obtained target data corresponding to the data attribute according to the data conversion rule to obtain the association value corresponding to the product to-be-predicted index.

In one embodiment, the predicted value obtaining module is configured to obtain a timestamp of the session message, and determine the target time period according to the timestamp and a reference window length of the linear regression model; acquiring a numerical attribute value corresponding to a product to-be-predicted index in a target time period, and inputting an association value and the numerical attribute value corresponding to the product to-be-predicted index in the target time period into a linear regression model to train the linear regression model; and inputting the correlation value of the time stamp corresponding to the time into a trained linear regression model, and outputting the numerical attribute predicted value of the index to be predicted of the product by the trained linear regression model.

In one embodiment, the session message processing apparatus further includes a reference regression window length obtaining module, where the reference regression window length module is configured to: acquiring a numerical attribute value and an associated value corresponding to a product to-be-predicted index in a first time period; acquiring a plurality of different pre-selected regression window lengths, and determining a second time period according to each pre-selected regression window length; respectively constructing a preselected linear regression model corresponding to the length of each preselected regression window according to the numerical attribute value and the associated value corresponding to the product to be predicted index in each second time period; respectively inputting the correlation values of the first time period into a preselected linear regression model corresponding to each preselected window length to obtain numerical attribute predicted values of a plurality of product to-be-predicted indexes in the first time period; and respectively calculating the error value between the numerical attribute predicted value and the numerical attribute value obtained through the pre-selected linear regression model corresponding to each pre-selected window length, and taking the pre-selected regression window length corresponding to the numerical attribute predicted value with the minimum error of the numerical attribute value as the reference regression window length.

In one embodiment, the reference regression window length module is specifically configured to obtain an adjustment coefficient of the reference window length; adjusting the length of the reference regression window according to the adjustment coefficient to obtain the optimal regression window length; the optimal regression window length is determined as the reference window length.

In one embodiment, the session message acquisition module is specifically configured to match the target text information with a known product identifier; if the target text information is matched with the known product identifier, determining the product identifier of the known product identifier as the product identifier corresponding to the target text information.

For specific limitations of the processing apparatus for session information, reference may be made to the above limitation of the processing method for session information, and no further description is given here. The respective modules in the processing apparatus of session information described above may be implemented in whole or in part by software, hardware, or a combination thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.

In one embodiment, a computer device is provided, which may be a server, the internal structure of which may be as shown in fig. 5. The computer device includes a processor, a memory, a network interface, and a database connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database of the computer equipment is used for storing data to be predicted, numerical attributes of different products and the like. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program, when executed by a processor, implements a method of processing session information.

It will be appreciated by those skilled in the art that the structure shown in fig. 5 is merely a block diagram of some of the structures associated with the present application and is not limiting of the computer device to which the present application may be applied, and that a particular computer device may include more or fewer components than shown, or may combine certain components, or have a different arrangement of components.

In one embodiment, a computer device is provided comprising a memory storing a computer program and a processor that when executing the computer program performs the steps of:

acquiring a data conversion instruction, and converting the numerical resource data into an associated value corresponding to the index to be predicted of the product according to the data conversion instruction;

and generating response information corresponding to the session message according to the numerical attribute predicted value in a preset format, and displaying the response information through a session interface.

In one embodiment, the processor executes the computer program to implement the step of converting the numerical resource data into the associated value corresponding to the index to be predicted of the product according to the data conversion instruction, and specifically includes the following steps: extracting a data conversion rule in a preset data conversion instruction; the data conversion rule is a conversion rule between the numerical resource data and the associated value; acquiring data attributes in the data conversion rule, and acquiring target data corresponding to the data attributes from the numerical resource data; and converting the obtained target data corresponding to the data attribute according to the data conversion rule to obtain the association value corresponding to the product to-be-predicted index.

In one embodiment, when the processor executes the computer program to input the association value corresponding to the product to-be-predicted index into the linear regression model to obtain the numerical attribute predicted value of the product to-be-predicted index, the following steps are specifically implemented: acquiring a time stamp of the session message, and determining a target time period according to the time stamp and the reference window length of the linear regression model; acquiring a numerical attribute value corresponding to a product to-be-predicted index in a target time period, and inputting an association value and the numerical attribute value corresponding to the product to-be-predicted index in the target time period into a linear regression model to train the linear regression model; and inputting the correlation value of the time stamp corresponding to the time into a trained linear regression model, and outputting the numerical attribute predicted value of the index to be predicted of the product by the trained linear regression model.

In one embodiment, the processor when executing the computer program further performs the steps of: acquiring a numerical attribute value and an associated value corresponding to a product to-be-predicted index in a first time period; acquiring a plurality of different pre-selected regression window lengths, and determining a second time period according to each pre-selected regression window length; respectively constructing a preselected linear regression model corresponding to the length of each preselected regression window according to the numerical attribute value and the associated value corresponding to the product to be predicted index in each second time period; respectively inputting the correlation values of the first time period into a preselected linear regression model corresponding to each preselected window length to obtain numerical attribute predicted values of a plurality of product to-be-predicted indexes in the first time period; and respectively calculating the error value between the numerical attribute predicted value and the numerical attribute value obtained through the pre-selected linear regression model corresponding to each pre-selected window length, and taking the pre-selected regression window length corresponding to the numerical attribute predicted value with the minimum error of the numerical attribute value as the reference regression window length.

In one embodiment, when the processor executes the computer program to implement the step of taking the preselected regression window length corresponding to the numerical attribute predicted value with the smallest error of the numerical attribute values as the reference regression window length, the following steps are specifically implemented: acquiring an adjustment coefficient of the length of a reference window; adjusting the length of the reference regression window according to the adjustment coefficient to obtain the optimal regression window length; the optimal regression window length is determined as the reference window length.

In one embodiment, when the processor executes the computer program to realize the step of obtaining the product identification code and the product to-be-predicted index corresponding to the target text information, the specific steps are as follows: matching the target text information with the known product identifier; if the target text information is matched with the known product identifier, determining the product identifier of the known product identifier as the product identifier corresponding to the target text information.

In one embodiment, a computer readable storage medium is provided having a computer program stored thereon, which when executed by a processor, performs the steps of:

In one embodiment, when the computer program is executed by the processor to implement the step of converting the numerical resource data into the associated value corresponding to the index to be predicted of the product according to the data conversion instruction, the following steps are specifically implemented: extracting a data conversion rule in a preset data conversion instruction; the data conversion rule is a conversion rule between the numerical resource data and the associated value; acquiring data attributes in the data conversion rule, and acquiring target data corresponding to the data attributes from the numerical resource data; and converting the obtained target data corresponding to the data attribute according to the data conversion rule to obtain the association value corresponding to the product to-be-predicted index.

In one embodiment, the computer program is executed by the processor to implement the step of inputting the association value corresponding to the product to be predicted index into the linear regression model to obtain the numerical attribute predicted value of the product to be predicted index, and specifically includes the following steps: acquiring a time stamp of the session message, and determining a target time period according to the time stamp and the reference window length of the linear regression model; acquiring a numerical attribute value corresponding to a product to-be-predicted index in a target time period, and inputting an association value and the numerical attribute value corresponding to the product to-be-predicted index in the target time period into a linear regression model to train the linear regression model; and inputting the correlation value of the time stamp corresponding to the time into a trained linear regression model, and outputting the numerical attribute predicted value of the index to be predicted of the product by the trained linear regression model.

In one embodiment, the computer program when executed by the processor further performs the steps of: acquiring a numerical attribute value and an associated value corresponding to a product to-be-predicted index in a first time period; acquiring a plurality of different pre-selected regression window lengths, and determining a second time period according to each pre-selected regression window length; respectively constructing a preselected linear regression model corresponding to the length of each preselected regression window according to the numerical attribute value and the associated value corresponding to the product to be predicted index in each second time period; respectively inputting the correlation values of the first time period into a preselected linear regression model corresponding to each preselected window length to obtain numerical attribute predicted values of a plurality of product to-be-predicted indexes in the first time period; and respectively calculating the error value between the numerical attribute predicted value and the numerical attribute value obtained through the pre-selected linear regression model corresponding to each pre-selected window length, and taking the pre-selected regression window length corresponding to the numerical attribute predicted value with the minimum error of the numerical attribute value as the reference regression window length.

In one embodiment, the computer program is executed by the processor to implement the step of taking as a reference regression window length a preselected regression window length corresponding to a numerical attribute predictor having a minimum error in the numerical attribute value, the steps of: acquiring an adjustment coefficient of the length of a reference window; adjusting the length of the reference regression window according to the adjustment coefficient to obtain the optimal regression window length; the optimal regression window length is determined as the reference window length.

In one embodiment, when the computer program is executed by the processor to implement the step of obtaining the product identification code and the product to-be-predicted index corresponding to the target text information, the following steps are specifically implemented: matching the target text information with the known product identifier; if the target text information is matched with the known product identifier, determining the product identifier of the known product identifier as the product identifier corresponding to the target text information.

Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the various embodiments provided herein may include non-volatile and/or volatile memory. The nonvolatile memory can include Read Only Memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link DRAM (SLDRAM), memory bus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), among others.

The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.

The above examples merely represent a few embodiments of the present application, which are described in more detail and are not to be construed as limiting the scope of the invention. It should be noted that it would be apparent to those skilled in the art that various modifications and improvements could be made without departing from the spirit of the present application, which would be within the scope of the present application. Accordingly, the scope of protection of the present application is to be determined by the claims appended hereto.

Claims

1. A method for processing session information, the method comprising:

generating response information corresponding to the session message according to the numerical attribute predicted value and a preset format, and displaying the response information through the session interface;

the step of inputting the association value corresponding to the product to-be-predicted index into a linear regression model to obtain the numerical attribute predicted value of the product to-be-predicted index comprises the following steps:

acquiring a time stamp of the session message, and determining a target time period according to the time stamp and the reference regression window length of the linear regression model;

inputting the correlation value of the time corresponding to the time stamp into a trained linear regression model, and outputting a numerical attribute predicted value of a product to be predicted index by the trained linear regression model;

Before the step of obtaining the timestamp of the session message, further comprises:

respectively inputting the association values of the first time period into preselected linear regression models corresponding to the lengths of the preselected regression windows to obtain numerical attribute predicted values of a plurality of product to-be-predicted indexes in the first time period;

and respectively calculating error values between the numerical attribute predicted values and the numerical attribute values, which are obtained through the pre-selected linear regression models corresponding to the pre-selected regression window lengths, and taking the pre-selected regression window length corresponding to the numerical attribute predicted value with the smallest error of the numerical attribute values as a reference regression window length.

2. The method according to claim 1, wherein the step of converting the numerical resource data into the associated value corresponding to the product to-be-predicted index according to the data conversion instruction includes:

Extracting a data conversion rule in the data conversion instruction; the data conversion rule is a conversion rule between the numerical resource data and the associated value;

3. The method of claim 1, wherein the step of taking as a reference regression window length a preselected regression window length corresponding to a numerical attribute predictor having a smallest error of the numerical attribute value, comprises:

acquiring an adjustment coefficient of the length of the reference regression window;

and determining the optimal regression window length as a reference regression window length.

4. The method of claim 1, wherein the step of obtaining the product identification code and the product to-be-predicted index corresponding to the target text information comprises:

Matching the target text information with a known product identifier;

5. A processing apparatus for session information, the apparatus comprising:

the response information generation module is used for generating response information corresponding to the session message according to the numerical attribute predicted value and a preset format, and displaying the response information through the session interface;

The predicted value acquisition module is further used for acquiring a time stamp of the session message, and determining a target time period according to the time stamp and the reference regression window length of the linear regression model; acquiring a numerical attribute value corresponding to the product to-be-predicted index in a target time period, and inputting an association value corresponding to the product to-be-predicted index in the target time period and the numerical attribute value into a linear regression model to train the linear regression model; inputting the correlation value of the time corresponding to the time stamp into a trained linear regression model, and outputting a numerical attribute predicted value of a product to be predicted index by the trained linear regression model;

the device further comprises a reference regression window length acquisition module, a reference regression window length acquisition module and a correlation module, wherein the reference regression window length acquisition module is used for acquiring a numerical attribute value and a correlation value corresponding to the product to-be-predicted index in a first time period; acquiring a plurality of different pre-selected regression window lengths, and determining a second time period according to each pre-selected regression window length; respectively constructing a preselected linear regression model corresponding to the length of each preselected regression window according to the numerical attribute value and the associated value corresponding to the product to be predicted index in each second time period; respectively inputting the association values of the first time period into preselected linear regression models corresponding to the lengths of the preselected regression windows to obtain numerical attribute predicted values of a plurality of product to-be-predicted indexes in the first time period; and respectively calculating error values between the numerical attribute predicted values and the numerical attribute values, which are obtained through the pre-selected linear regression models corresponding to the pre-selected regression window lengths, and taking the pre-selected regression window length corresponding to the numerical attribute predicted value with the smallest error of the numerical attribute values as a reference regression window length.

6. The apparatus of claim 5, wherein the association value acquisition module is specifically configured to: extracting a data conversion rule in the data conversion instruction; the data conversion rule is a conversion rule between the numerical resource data and the associated value; acquiring data attributes in the data conversion rule, and acquiring target data corresponding to the data attributes from the numerical resource data; and converting the obtained target data corresponding to the data attribute according to the data conversion rule to obtain the association value corresponding to the product to-be-predicted index.

7. The apparatus of claim 5, wherein the reference regression window length acquisition module is configured to: acquiring an adjustment coefficient of the length of the reference regression window; adjusting the length of the reference regression window according to the adjustment coefficient to obtain the optimal regression window length; and determining the optimal regression window length as a reference regression window length.

8. The apparatus of claim 5, wherein the session message acquisition module is specifically configured to: matching the target text information with a known product identifier; and if the target text information is matched with the known product identifier, determining the product identifier of the known product identifier as the product identifier corresponding to the target text information.

9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any of claims 1 to 4 when the computer program is executed.

10. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 4.