WO2018091117A1

WO2018091117A1 - System and method for ascertaining a degree of match between electronic documents

Info

Publication number: WO2018091117A1
Application number: PCT/EP2016/078308
Authority: WO
Inventors: Peter Kolb
Original assignee: Logon Consulting Gmbh & Co. Kgaa
Priority date: 2016-11-21
Filing date: 2016-11-21
Publication date: 2018-05-24

Abstract

The invention relates to a method and a system for ascertaining a degree of match between a number of first electronic documents and at least one second electronic document, wherein the method comprises at least the following steps: a) a first step, in which the electronic documents are conditioned and lists of terms are derived from the conditioned documents, wherein the respective list of terms is used to convert each of the conditioned documents into a numerical sequence, so that the numerical sequences represent electronic documents, b) a second step, in which profiles are created from the numerical sequences, and c) a third step, in which the profiles are compared with one another in order to ascertain disparities between the number of first electronic documents and the at least one second electronic document, wherein the disparities indicate the degree of match between the number of first electronic documents and the at least one second electronic document.

Description

System and method for determining a degree of correspondence between electronic documents

Field of the invention

The invention relates to a system and a method for determining a degree of agreement between electronic documents, in particular between a number of first electronic documents and at least one second electronic document to determine how well the number of first electronic documents content to the at least one second electronic content, whereby content and its meaning as well as the background of the content are determined in order to determine the degree of correspondence.

Background of the Invention The invention is described in terms of a recruiting process in which application documents must be compared with a job advertisement to determine how well an application matches a job advertisement. However, the invention is not limited to this and can also be applied to other areas in which electronic documents have to be compared in terms of content.

Due to the accelerating technological development, there is also an ever faster digitization of the market and processes in the field of personnel recruitment. So more and more companies are trying that To digitize the recruitment process, applicants are increasingly demanding that applicants submit their applications electronically.

Digitization is also increasing the reach of job advertisements steadily, allowing more and more potential candidates to be reached in ever shorter time. Candidates can increasingly apply more easily, for example by e-mail or simply by providing a link to a social media profile. As a result, there is a strong increase in online applications, which leads to a strong increase in the number of unsuitable applications.

Companies are thus faced with the challenge of quickly and reliably filtering out suitable and interesting applications for a job advertisement from a large number of online applications in order to get in touch with the potential candidates. As a result, companies must compare and / or compare applications with a job advertisement as quickly as possible.

The conventional view of the applications does not work here for cost and time especially reasons. In terms of response times, digitization leads candidates to expect them to respond as promptly as possible, preferably in real time. The reality is far from that. For example, for the German DAX ^® 50 companies, the average processing time from online applications to feedback is over 10 weeks, which in the worst case means that a suitable candidate is no longer available after feedback from the company ,

Another problem is that very few companies have enough experience to judge each application safely. Most companies are simply overwhelmed by this, especially since the application documents usually do not contain the necessary information that the staff member would want to assess the application. As a result of the long processing cycles and, in some cases, little valuable feedback, job applicants are increasingly dissatisfied with the application process, especially as the gap between the digital world in the private environment of the applicant and the analog, sometimes ancient world of HR departments seems to be growing.

The problem of timely and secure evaluation of a steadily increasing number of online applications, companies usually try to get a handle on two solutions:

1. Appointment of external service providers who process and evaluate the application.

This solution does not work in two ways:

- First, the service providers are also working analogously, which only increases the applicants' dissatisfaction. In addition, there is now one step more and direct contact with the future employer is made more difficult.

- Secondly, this solution represents a significant cost factor. German companies spend EUR 1.8 billion per year on this type of service.

2. Companies require applicants to complete specific, standardized application forms in order to receive structured and digitally evaluable applications. However, this solution leads to two other problems that make this solution useless:

- First, it moves the problem from the company to the candidate, who now has the effort to complete the form. Already leaning today 10% of the applicants do not submit any application forms and do not apply at all. 40 more percent will drop the application if the application takes more than 15 minutes to complete. By using application forms, companies lose up to 50% of their applicants, which is a huge problem in times of increasing employee shortages.

In addition, a mobile application using forms is not meaningful, since the filling of such an important form on the smartphone is impractical and the voice control for privacy reasons, at least in the public areas (such as subway, etc.) is omitted as input medium.

Furthermore, when applying by means of standardized application forms, information that may be of relevance to the selection of an applicant, for example because the application form does not provide an input field for this information, may not be considered. However, if such information is sent as an attachment to an application form, for example, there is again the problem that the attachment must be viewed manually, which largely nullifies the actual advantage of a structured form. Secondly, the use of application forms in times of digitization, social profiles and big data represents a huge loss of image in terms of the employer brand. The lack of digitization in recruitment is one-to-one transferred to a lack of digitization of the company and its products , As a result, the company loses not only applicants, but also customers.

Object of the invention The object of the present invention is therefore to provide solutions which make it possible to compare a number of first electronic documents, in particular at least partially unstructured documents, with at least one second electronic document, in particular at least partially unstructured document, and to provide a value for a match In particular, for an application process, the above-mentioned disadvantages are avoided, so that even with unstructured documents a timely, possibly in real time, feedback is guaranteed. Furthermore, an object of the invention is to considerably simplify the application process for the applicant.

Inventive Solution According to the invention, this object is achieved by a method and a system according to the independent patent claims. Advantageous developments and refinements of the invention are specified in the respective dependent claims. Accordingly, a method is provided for determining a degree of agreement between a number of first electronic documents and at least one second electronic document in a server device, wherein the number of first electronic documents and the at least one second electronic document are transmitted to the server device via a communication network and there at least for the determination of the degree of agreement is stored, the method comprising at least the following steps:

a) a first step, in which the number of first electronic documents and the at least one second electronic document are processed and from the processed documents at least a first list of terms associated with the number of first electronic documents, and at least one second term list, which is assigned to the at least one second electronic document, are derived, with the respective list of terms the edited documents are each converted into a sequence of numbers, so that a first sequence of numbers the number of first electronic documents and a second sequence of numbers the at least one second electronic document represent,

a second step in which

at least a first profile from the first sequence of numbers, and

from the second sequence of numbers at least a second profile

be created, and

a third step in which the at least one first profile is compared with the at least one second profile to determine deviations between the number of first electronic documents and the at least one second electronic document, wherein the deviations between the number of first electronic documents and indicate the at least one second electronic document.

It is advantageous if in the first step a)

the first and second electronic documents are each converted into plain text documents and unifying or removed formatting marks,

the pure text documents are each structured into content-related document areas, and

the document areas are read out individually and predetermined terms are determined for each read-out document area, wherein each of the determined terms is evaluated, wherein the evaluated terms are stored in the first list of terms and in the second list of terms and the term list is assigned to the respective document area. Remotely, it is advantageous if the structuring of the pure text documents is carried out on the basis of subject areas stored for the first and second electronic documents in a respective topic table, and / or

the predetermined terms are each stored in match tables.

The evaluation of the determined terms can be done in such a way that

each term has been categorized into one or more classes of terms (Krit Kat), depending on the area of the document in which the term was identified, and

a level is determined for each term determined, the level indicating the relevance of the term, and a weighting (LvL) being assigned to the term according to the level.

In this case, it can be advantageous if each term in the term class is assigned a number of criteria which characterize the respective term and wherein each term determined is evaluated separately for each term class in which it is classified, whereby the rating is carried out by in that one or more criteria of the terms in the conceptual class is assigned a numerical value, the numerical values for each criterion being stored in the first or second list of terms for the respective term.

In the second step b) can

For each document area, at least one capability profile is created, each capability profile being created on the basis of a base profile and the list of terms associated with the respective document area, each one being created

Base profile is associated with a basic term list and wherein only those base profiles are taken into account, which have at least one basic term, which is also included in the list of terms associated with the respective document area, and wherein

- For each document area an area profile is generated from the capability profiles created for the respective document area. It has proven to be advantageous if, when generating a capability profile, those criteria and the associated numerical values are selected from the list of terms which are assigned to the respective term according to the base profile.

When generating an area profile, a weighted average of all the values of the respective criteria can be determined for each criterion of the capability profiles generated for the respective document area, wherein the weighted average values of all criteria form the area profile for a document area.

It is also advantageous if, using the area profiles, the first profile is produced, which comprises a first partial profile and a second partial profile, the first partial profile comprising the criteria and associated values of the area profiles, and

- The second sub-profile comprises trend values for the criteria covered by the first sub-profile.

The associated values may be selected from the group comprising minimum value of a criterion across all range profiles, maximum value of a criterion across all range profiles, average value of a criterion across all range profiles, maximum quality level across all terms, off-target accuracy, and combinations thereof.

The second subprofile can be generated by

- all area profiles are sorted according to a predetermined criterion, preferably chronologically ascending, and

a trend value is determined for each criterion.

It may also be advantageous if second profile is generated using the area profiles, wherein an average profile is formed from the values of the respective criteria of the area profiles,

from the values of the criteria associated with the terms associated with a predetermined property, a weighted average is calculated for each criterion, the respective weighted averages forming a mandatory profile, and

summarizing for each criterion the values of the average profile and the must profile to create the second profile. In the third step c), the comparison between the two profiles can be carried out by comparing the individual criteria of the first profile with the respective criteria of the second profile.

The comparison can be carried out in an advantageous manner by subtracting the individual values of the respective criteria.

A system is also provided for determining a degree of agreement between a number of first electronic documents and at least one second electronic document, the system having a server device which can be coupled via a communication network to a first client device and to a second client device, wherein the electronic documents are transferable to the server device via the communication network, and wherein the server device is adapted to carry out the method according to the invention.

Brief description of the figures

Details and features, in particular advantageous embodiment and further developments of the invention will become apparent from the following description taken in conjunction with the drawings. It shows: 1 shows an example of a first electronic document, which is present in an unstructured form and represents a typical application document;

Fig. 2 is an example of a plain text document generated according to the invention from the electronic document shown in Fig. 1;

3 shows an example of relevant terms of a section of the text document shown in FIG. 2; and

4 shows a system according to the invention which is adapted to the invention

Perform procedure.

Detailed description of the invention

The solution according to the invention is based on the concept that a number of first electronic documents, such as application documents, which may be in unstructured form, are compared with at least one second electronic document, such as a job advertisement, which may also be in unstructured form, namely such that both for the first documents and for the second document a respective metric is calculated, which can be compared with each other, so that the comparison of the two metrics provides information about how well the first documents match the second document. This allows us to make a fully automatic comparison, which in turn makes it possible to promptly submit feedback on an application. With regard to a recruiting or application process, the invention provides the following advantages:

The applicant simply applies online or mobile, no matter where he is, by uploading a CV (for example from the cloud) or with a SocialMedia profile (eg XING, LinkedIn) or by uploading any other application documents, such as certificates, cover letters or similar. The process of "applying yourself" is considerably simplified. The applicant no longer has to worry about formatting, structuring the documents, etc.

- At the moment of sending the application, the application and the job offer will be read out, analyzed and interpreted by the invention online and in real time.

- Based on this, the application is evaluated and an estimate of the accuracy of the passport is drawn up.

- Within a few seconds of submitting the application, the evaluation is accurate and reliable for both the employer and the applicant.

The inventive method is essentially divided into three sections or main steps, namely

- Step 1, the so-called Information Gathering, in which the number of first electronic documents (eg a number of application documents, eg cover letters, certificates, etc.) and the at least one second electronic document (eg a job advertisement) are processed. From the processed documents at least a first term list is generated, which is assigned to the number of first electronic documents. Furthermore, at least one second list of terms is generated from the edited documents, which is assigned to the at least one second electronic document. Using the respective list of terms, processed documents are converted into a sequence of numbers or a sequence of numbers is generated in each case. The number sequences here represent the respective documents and can therefore be regarded as a metric for the respective document. - Step 2, the so-called profiling, in which from the first sequence of numbers at least a first profile (for example, an applicant profile) and from the second Number sequence at least a second profile (for example, a job profile) are created, and

- Step 3, the so-called matching, in which the at least one first profile is compared with the at least one second profile to determine deviations between the number of first electronic documents and the at least one second electronic document, the deviations a degree of agreement between the number first electronic documents and the at least one second electronic document. The degree of agreement here indicates how well the first documents match the at least one second. For the application process, this means that the degree of correspondence indicates how well an application matches a job advertisement. Step 1: Information Gathering

First, the first documents, such as the application documents, and the second documents, such as a job advertisement, each converted into pure text documents. An example of a first document in the form of a standard application is shown in FIG. The first document can be any document or come from any source. Thus, for example, a PDF document, an HTML document or website or an online profile of an applicant can be made available as the first document. A pure text document in the sense of the present invention is an electronic document which exclusively comprises letters, numbers and predetermined structuring symbols (eg *, $, §, etc.). Formatting characters are standardized and, if necessary, replaced by predetermined structuring symbols. An example of a plain text document generated from the document shown in FIG. 1 is shown in FIG. Next, the pure text documents are structured into content-related or thematically related areas and, if dates are available, these areas are sorted by time. These related areas are also called document areas below. The possible subject areas in which the pure text documents can be structured or structured are stored in so-called "title tables", for example in a database. Examples of such topics include education, work experience, hobby, etc.

After structuring the pure text documents, these are read out in regions and for each read document area predetermined terms are determined, which are relevant in the context in which the electronic documents are processed. In the context of a recruitment process, the predefined terms can be, for example, recruitment-relevant terms. These predetermined terms are preferably stored in so-called "match tables", for example in a database.

Semantic methods can also be used to determine those terms in plain text documents that are not stored in the match tables. The terms thus determined are then preferably stored in the match tables.

In one embodiment of the invention, such terms can be stored in the match tables, which should be ignored in the term determination.

An example of the relevant terms found in the text document shown in FIG. 2 (in this case recruiting-relevant terms) is shown in FIG.

Next, each of the found relevant terms is automatically evaluated: Depending on the document area in which the relevant term was found (for example, education, work experience, hobby), it is classified into one or more term classes or category of terms (Krit Kat). The classification into the term classes can be accomplished by means of further match tables and methods provided for this purpose.

For each relevant term, a so-called "level" is determined. That is, it determines how well or intensively this term has been used in, for example, the first electronic documents (such as application documents) or how important that term is for the second electronic document (such as a vacancy notice). According to the level, a weighting (LvL) is then assigned to each relevant term.

Each of the above term classes may be assigned any number of criteria (Crit1 to CritN) that characterize that term class in the particular context in which the electronic documents are processed (such as from a recruitment perspective). In the application process, these criteria may include, for example, hard skills, soft skills, skills or the like.

Each found relevant term is evaluated with respect to all criteria of the term class with the corresponding values of the term class, i. A term is assigned the criteria of the term class to which it is assigned, to which in turn a value of the term class, preferably a natural positive or negative number, is assigned, whereby the values of the individual criteria within a conceptual class or term category are always the same.

In addition, each term is evaluated separately for each term class to which it is assigned. The same term can thus be classified in several categories of terms and is evaluated there independently of the other term classes. The following is an example of an evaluation of relevant terms for sales jobs and distribution candidates:

For example, relevant terms for sales jobs and sales candidates are evaluated using the following criteria. For each ski area (for example, sales, IT, accounting, etc.) you can define your own criteria model, which is uniform for all terms in this ski area. The criteria model itself can also be stored in a database. A criteria model may include n criteria, where n is preferably arbitrary. In a term class, only those criteria that can be evaluated are evaluated, non-evaluable criteria in a term class are ignored. Exemplary criteria model for the ski area "Sales":

- Kritl field service share

- Crit2_ increase in sales

- Crit3_New customer share

- Crit4_complexity of the decision maker structure

- Crit5_complexity of the market

- Critical complexity of the product

- Critical part of management responsibility

- CritAB completion orientation

- KritKO customer orientation

- CritPL problem solving

- CritTO team orientation

KritN The evaluation of the individual criteria preferably takes place by means of numerical values. It is advantageous here if the criteria are selected such that a "from-to" evaluation is made possible:

- If the criteria are rated with numbers from 1 to X, the assessment shows how well the respective criterion is pronounced in terms of this term (absolute criterion).

- If the criteria are rated with numbers from -X to + X, the assessment shows how strongly the terms of this term are below or above the average for the respective criterion (relative criterion).

Then each relevant term is cached for each term class to which it has been assigned with the following attributes or attribute values in a list or table:

- term identifier

- Conceptual class

- Value for Critl to CritN with the respective numerical value (this results in an N-digit number, a so-called profile number)

- membership of the date range (from ... to ...) and if necessary duration in months / weeks, etc.

- Belonging to the document area

- Ski area

- Level

- Position at which the term was found in the respective document

The result of the automatic evaluation is then a list of terms, with values for the aforementioned attributes. This means that the pure text documents have now been converted into a mathematically editable and evaluable sequence of numbers, so that a computer-assisted comparison can be carried out with special care and efficiently. In addition to the list of terms, the sections of the pure text documents belonging to the individual document areas and date ranges can be temporarily stored, for example, in order to enable a later display of the corresponding text areas.

The information gathering can be carried out by a so-called "information gathering engine". Step 2: Profiling

The profiling is divided into three substeps, namely:

1. Create an area profile,

2. Create the first profile (e.g., an applicant profile), and

3. Create the second profile (e.g., a job profile)

The area profile is generated from the number sequences of the previously created list of terms.

In this case, skill specific individual profiles (also called ability profiles) are initially created for each ski area (for example, sales, IT, accounting, etc.). For each ski area only those entries of the list of terms are used which are assigned to the respective ski area (according to the above-mentioned attribute "ski area"). This means that several skill-specific individual profiles can be created for each ski area.

In addition, a separate profile (area profile) is created for each document area (eg training, work experience, hobby), for example the profiles Training 1, Training 2, Job 1, Job 2, Hobby for applicants or the profiles Requirements, Tasks and Experience in job advertisements. Are for the respective Document Range Dates available, the profile is assigned the lowest date. The date can later be used for chronological sorting of the range profiles. To create the area profiles, the weighted average is calculated from all the values of the respective criterion for each criterion (the criteria of the list mentioned above). The weighting takes place according to the level and duration assigned to the criterion value, if available (see above attributes of the term list). In the area of recruiting, the weighted average represents the value of a competency in one area of the CV or the job advertisement. All the weighted average of a document range together form the range profile.

The weighted average of a criterion (below for criterion criterion) for a specific document area is calculated as follows:

((Critl (1) * L1 (1) * D1 (1)) + ··· + (critl (n) * Ll (n) * Dl (n)))

Σ Lb * Σ Db with

Kritl avg (b) = average for Kritl in the range b

Kritl (l) = first value of criterion Kritl

Critl (n) = nth value of criterion Kritl

Ll (l) = Level of the first value of criterion Kritl

Ll (n) = level of the nth value of criterion Kritl

Dl (l) = duration of the first value of criterion Kritl

Dl (n) = duration of the nth value of criterion Kritl

Lb = all level information in the area b

Db = all duration information in the range b The range profile for the range b therefore consists of the values Kritlavg (b) ... KritNavg (b), Dl ... DN, LI ... LN, the range ID, Min (date), Max (date), and Duration (Max) -Min (Date)). In the next step, the first profile (eg the applicant profile) is created.

For a candidate profile, a first sub-profile (also called experience profile) and a second sub-profile (also called target profile) are created from the area profiles previously created for the first electronic documents (for example application documents). The experience profile summarizes the competences of the candidate from the past. The target profile creates a prognosis for the candidate's competencies. For both profiles, i. For the experience profile and for the target profile, it is first checked for each criterion whether sufficient terms have been evaluated in order to be able to make a reliable statement with regard to the respective criterion (which indeed represents a competence of an applicant). It has proved to be advantageous if at least three terms have been evaluated for a criterion in order to be able to output a profile value for this criterion.

To create the experience profile, the following profile values are determined for each sufficiently evaluated criterion: - Min (Critl) ... Min (KritN)

- Max (Crit 1) ... Max (CritN)

- AVG (Critl) ... AVG (CritN)

- LvL (Kritl) ... LvL (KritN)

- PQ (Kritl) ... PQ (KritN) with Min = Minimum Value of the respective criterion over all area profiles

Max = maximum value of the respective criterion over all area profiles

AVG = average value of the respective criterion over all area profiles LvL = maximum quality level over all terms

PQ = profiling quality

To create the destination profile, first all area profiles are sorted chronologically (preferably in ascending order), with area profiles without dates being set to the beginning (or the end, depending on the sorting order).

Then, using a trend determination (for example, by means of a linear regression model), a trend value is determined for each criterion as a forecast for the future, so that the target profile results in the values

- trnd (Kritl) ... trnd (KritN), with trnd = trend value of the respective criterion across all area profiles.

For each available ski area (for example, sales, IT, accounting, etc.), the experience profile and the target profile are combined into an n-digit sequence of numbers, which is also called an applicant skill profile. Thus, several candidate skill profiles can be generated or be present.

The first profile (applicant profile) is then formed from possibly several candidate skill profiles. In the next step, the second profile (eg the second profile) is then generated from the area profiles generated for the second electronic documents (eg job advertisements) Job profile). The profiling is done separately for each ski area, so that there may be several profiles for a job advertisement. From the values of the criteria of the area profiles, an average profile (DP) is formed by means of a simple average.

From all the values of the criteria for the terms with "must-level", a so-called must profile (MP) is formed by a weighted average. The values of the average profile and the required profile are then combined for each ski area into the second profile (job profile), which then contains the following values:

DP Critl, MP Rritl ... DP CritN, MP CritN

The profiling can be done by a so-called "profiling engine".

Step 3: Matching

In this step, the previously created profiles are compared with each other. Due to the structure of the first profiles (applicant profiles) and the second profiles (job profiles), individual profiles can be compared in any combination: experience profile, target profile, must profile, average profile etc.

The evaluation is carried out here by comparing the individual criteria with one another, for example by simple subtraction, as shown below by the example of the comparison of an average profile with a target profile:

DP Kritl AVG Kritl AKrit 1 (DP AVG)

DP Krit2 AVG Kritl AKrit2 (DP_AVG)

DP CritN AVG CritNI AKritN (DP AVG) Depending on the requirements, however, other profile values of the average profile and the target profile can also be compared. Thus, in addition to the above example, the average profile of the job advertisement can additionally be compared with a trend profile of the applicant.

This makes it possible, for example, to determine the deviation between the applicant and the job application or between the application and the job advertisement for each criterion. The sum of all deviations represents the fit accuracy or match level (ML) of the applicant and the job advertisement.

The respective deviations can be summarized by means of report generators in a so-called match report and evaluated for providing an overall decision on the fit accuracy with a neural network.

The number of available evaluated criteria in relation to the total number of assessable criteria of a criteria model represents match accuracy (MG) and is thus an indicator of the reliability of the rating.

Using a code generator, text modules, icons, graphics or process steps can be generated according to the different deviations, which can be provided to the applicant or the company as feedback. Thus, feedback on an application can be created fully automatically, so that overall, feedback in real time is possible.

Of course, the criteria do not have to be compared individually. In one embodiment of the invention, criteria can be grouped together so that a comparison of individual criteria groups can be performed. The matching can be carried out by a so-called "Matching Engine".

The invention also provides a system adapted to carry out the method according to the invention. An example of such a system is shown in FIG.

The system here comprises a server device S which can be coupled or coupled via a communication network (for example the Internet) with at least one first client C1 and with at least one second client C2. The first client Cl can for example be assigned to an applicant B. For example, the second client C2 may be associated with a company U looking for new employees for a particular business area. By means of the client Cl, the applicant B can transfer one or more application documents to the server device S, where they can be stored in a database DB. Alternatively or additionally, the applicant B can also use the client Cl to transfer links L to one or more social media profiles to the server device S. The links L and / or the social media profiles referenced by the links can also be stored in the database DB.

By means of the client C2, the company U can transfer its job advertisement to the server device S, where it can be stored in the database DB.

After transferring the application documents Dl / Links L to the server device, the latter executes the above-mentioned method in order to automatically provide feedback on the applicant B's application, which indicates how well the application matches the job advertisement. The feedback can be sent to the client Cl be transferred. Such feedback can also be transferred to client C2 of company U immediately after the creation.

Here, the server device S preferably has an information gathering engine IE, a profiling engine PE and a matching engine ME, which are functionally coupled to each other or functionally interlock with each other. The information gathering engine IE is adapted to perform the above-mentioned step 1 of the method. The profiling engine PE is adapted to perform the above-mentioned step 2 of the method. The matching engine ME is adapted to perform the above-mentioned step 3 of the method.

Claims

claims

A method for determining a degree of agreement between a number of first electronic documents and at least one second electronic document in a server device, wherein the number of first electronic documents and the at least one second electronic document are transmitted via a communication network to the server device and there at least for determining the degree of agreement the method comprises at least the following steps: a) a first step in which the number of first electronic documents and the at least one second electronic document are prepared and from the processed documents at least a first list of terms associated with the number of first electronic documents is, and at least a second list of terms associated with the at least one second electronic document are derived, with the respective list of terms, the edited documents each i n is converted into a sequence of numbers such that a first sequence of numbers represents the number of first electronic documents and a second sequence of numbers represents the at least one second electronic document. b) a second step in which

from the first sequence of numbers at least a first profile, and

- from the second sequence of numbers at least a second profile

be created, and

c) a third step, in which the at least one first profile is compared with the at least one second profile to deviations between the number of first electronic documents and the at least determine a second electronic document, the deviations indicating the degree of correspondence between the number of first electronic documents and the at least one second electronic document.

Method according to the preceding claim, wherein in the first step a)

- the first and second electronic documents are each converted into plain text documents and unifying or removed formatting marks,

- the pure text documents are each structured into content-related document areas,

the document areas are read out individually and predetermined terms are determined for each read-out document area, wherein each of the determined terms is evaluated, wherein the evaluated terms are stored in the first list of terms and in the second list of terms and the term list is assigned to the respective document area.

A method according to the preceding claim, wherein

- The structuring of the pure text documents is performed on the basis of subject areas that are stored for the first and second electronic documents in a respective theme table, and / or

- The predetermined terms are stored in each match tables.

Method according to one of the preceding claims 2 or 3, wherein the evaluation of the determined terms is carried out such that

depending on the document area in which the term has been identified, each term determined is classified in one or more categories of terms (Krit Kat), and a level is determined for each term determined, the level indicating the relevance of the term, and a weighting (LvL) being assigned to the term according to the level.

Method according to the preceding claim, wherein each term in the term class is associated with a number of criteria characterizing the respective term and wherein each term determined is evaluated separately for each class of terms in which it is classified, the rating being by: one or more criteria of the terms in the conceptual class is assigned a numerical value, wherein the numerical values for each criterion in the first or second term list is stored to the respective term.

Method according to one of the preceding claims 2 to 5, wherein in the second step b)

- For each document area at least one capability profile is created, each capability profile is created based on a base profile and the respective document area associated list of terms, each base profile is assigned a basic term list and only those base profiles are considered, which have at least one basic term, which also is contained in the respective document area associated list of terms, and wherein

- an area profile is generated for each document area from the capability profiles created for the respective document area.

Method according to the preceding claim, wherein, when generating a capability profile, those criteria and the associated numerical values are selected from the list of terms which are assigned to the respective term according to the base profile.

8. The method according to claim 6, wherein, when generating an area profile for each criterion of the capability profiles generated for the respective document area, a weighted mean value is determined from all the values of the respective criteria, wherein the weighted average values of all criteria are the area profile for one Form document area.

Method according to one of claims 6 to 9, wherein using the area profiles, the first profile is generated, comprising a first part profile and a second part profile, wherein

- the first subprofile includes the criteria and associated values of the area profiles, and

- the second sub-profile comprises trend values for the criteria covered by the first sub-profile. 10. The method according to the preceding claim, wherein the associated values are selected from the group comprising minimum value of a criterion over all range profiles, maximum value of a criterion across all range profiles, average value of a criterion over all range profiles, maximum quality level across all terms, off-target accuracy, and combinations thereof ,

Method according to one of the two preceding claims, wherein the second sub-profile is generated by

- a trend value is determined for each criterion.

Method according to one of claims 6 to 11, wherein the second profile is generated using the area profiles, wherein

- an average profile is formed from the values of the respective criteria of the area profiles, from the values of the criteria associated with the terms to which a predetermined property is assigned, a weighted average for each criterion is calculated, the respective weighted averages forming a mandatory profile, and

- for each criterion, the values of the average profile and the required

Profiles to create the second profile.

13. The method according to any one of the preceding claims, wherein in the third step c) the comparison between the two profiles is performed by comparing the individual criteria of the first profile with the respective criteria of the second profile.

14. The method according to the preceding claim, wherein the comparison is carried out by subtracting the individual values of the respective criteria.

15. A system for determining a degree of correspondence between a number of first electronic documents and at least one second electronic document, the system having a server device (S) which can be coupled via a communication network to a first client device (Cl) and to a second client device (C2) wherein the electronic documents are transferable to the server device via the communication network, and wherein the server device is adapted to perform the method of any one of claims 1 to 14.