FR3144886A1 - Method and device for classifying vehicle repair support requests - Google Patents

Abstract

The present invention relates to a method and a device for classifying a request for repair of a vehicle component. For this purpose, the request data is processed (212) to be formatted. Information is extracted and/or generated (213) from this request. The data is then encoded (214) in a determined format to feed a classification model learned using an unsupervised learning method for classification (215) of the care request. Figure for abstract: Figure 2

Description

Method and device for classifying vehicle repair support requests

The present invention relates to methods and devices for classifying data representative of a request for repair of a vehicle component or organ, for example an automobile. The present invention also relates to a method and a device for monitoring the repair of a vehicle component or organ.

Technology background

Contemporary vehicles are made up of a large number of components or organs, each susceptible to breakdown or defect.

When a vehicle user is confronted with a defect or breakdown of one or more components of his vehicle, he contacts a specialized center to correct the defect or repair the vehicle. When the vehicle is under manufacturer's warranty, the correction of certain defects and the repair of certain breakdowns is covered by the manufacturer.

To obtain coverage for the correction or repair operations carried out, the specialized center sends a request or request for coverage of the work carried out to the vehicle manufacturer, such a request also being called a “credit request”.

To ensure proper processing of these requests, a category or class is associated with each request received, which makes it possible, for example, to determine or identify which parts of a type of vehicle suffer the most breakdowns in order to then make corrections or launch new developments to compensate for defects or breakdowns.

An incorrect classification thus leads to problems in monitoring support requests and correcting any breakdowns or defects observed. The sources of misclassification are numerous and include, for example, operator input problems.

Summary of the present invention

An object of the present invention is to solve at least one of the problems of the technological background described above.

Another object of the present invention is to improve the classification of vehicle component repair support requests.

According to a first aspect, the present invention relates to a method for classifying first data representative of a request for repair of at least one component of a first vehicle, the method being implemented by at least one processor , the process comprising the following steps:
- reception of the first data;
- processing of the first data to format the first data according to a determined structure;
- extraction of a first set of determined information from the first formatted data and generation of a second set of information from vehicle usage data;
- coding of the first and second information in a determined format;
- classification of the first data by feeding a classification model with the first and second coded information, the classification model having been learned in an unsupervised learning phase of a machine learning method;
- validation of a classification result from a set of analytical rules using a set of parameters obtained from databases storing data representative of defects and/or component repair processes associated with the vehicle;
- transmission of the validated result of the classification.

Classifying data from a classification model learned using a machine learning method reduces errors associated with manual processing of received support requests, thereby improving the classification of these support requests in charge. Learning in an unsupervised mode makes it possible to reduce data preparation and avoid biases linked to the classification of training data before training the classification model.

According to one variant, the learning phase comprises the following steps:
- obtaining, for each second vehicle of a set of second vehicles, second data representative of a request to cover the repair of at least one component of each second vehicle;
- processing of the second data to format said second data according to said determined structure;
- extraction of a third set of determined information from the second formatted data and generation of a fourth set of information from usage data of each second vehicle;
- coding of the third and fourth information in a determined format;
- unsupervised learning of the classification model from the third and fourth information coded in relation to the third data.

According to an additional variant, the unsupervised learning comprises a determination of a plurality of classes and a determination of a name for each class of the plurality of classes from the third and fourth information coded in relation to the third data.

According to another variant, the unsupervised learning further comprises a determination of a set of naming criteria for the plurality of classes from the third and fourth information coded in relation to the third data.

According to yet another variant, the learning phase further comprises a step of validating the classification model from third data representative of a set of requests for repair support for at least one component of each third vehicle d a set of third vehicles, fifth class representative data from the determined set of classes being associated with the fourth data.

According to an additional variant, the learning phase further comprises a step of semi-supervised learning of the classification model subsequent to the unsupervised learning, the semi-supervised learning being obtained as a function of the first and second associated coded information to a validated result of the classification obtained from the validation.

According to another variant, the processing of the first data comprises processing using an automatic natural language processing method.

According to an additional variant, the at least one processor implements a neural network.

According to a second aspect, the present invention relates to a device for classifying data representative of a request for repair of at least one vehicle component, the device comprising a memory associated with a processor configured for the implementation steps of the method according to the first aspect of the present invention.

According to a fourth aspect, the present invention relates to a computer program which comprises instructions adapted for the execution of the steps of the method according to the first aspect of the present invention, in particular when the computer program is executed by at least one processor.

Such a computer program can use any programming language, and be in the form of a source code, an object code, or an intermediate code between a source code and an object code, such as in partially compiled form, or in any other desirable form.

According to a fifth aspect, the present invention relates to a computer-readable recording medium on which is recorded a computer program comprising instructions for executing the steps of the method according to the first aspect of the present invention.

On the one hand, the recording medium can be any entity or device capable of storing the program. For example, the medium may comprise a storage means, such as a ROM memory, a CD-ROM or a ROM memory of the microelectronic circuit type, or even a magnetic recording means or a hard disk.

On the other hand, this recording medium can also be a transmissible medium such as an electrical or optical signal, such a signal being able to be conveyed via an electrical or optical cable, by conventional or terrestrial radio or by self-directed laser beam or by other ways. The computer program according to the present invention can in particular be downloaded onto an Internet type network.

Alternatively, the recording medium may be an integrated circuit in which the computer program is incorporated, the integrated circuit being adapted to execute or to be used in executing the method in question.

Brief description of the figures

Other characteristics and advantages of the present invention will emerge from the description of the particular and non-limiting examples of embodiment of the present invention below, with reference to the figures in Figures 1 to 4 attached, in which:

schematically illustrates a data communication network, according to a particular and non-limiting embodiment of the present invention;

schematically illustrates a process for classifying data representative of a request for repair support for at least one vehicle component implemented by a data processing device of the network of the , according to a particular and non-limiting embodiment of the present invention;

schematically illustrates a network device of the configured for the classification of data representative of a request for repair of at least one vehicle component, according to a particular and non-limiting example of embodiment of the present invention;

illustrates a flowchart of the different stages of a process for classifying data representative of a request for repair of at least one vehicle component implemented by the device of the , according to a particular and non-limiting embodiment of the present invention.

Description of the implementation examples

A method and a device for classifying data representative of a request for repair of at least one vehicle component will now be described in what follows with reference jointly to Figures 1 to 4. The same elements are identified with the same reference signs throughout the description which follows.

There schematically illustrates a data communication system or network 1, according to a particular and non-limiting embodiment of the present invention.

The system 1 comprises for example a data processing device 10 of the computer or laptop type. According to one variant, the device 10 corresponds to a mobile communication device, for example a smart phone (from the English “Smartphone”) or a tablet.

The device 10 is connected by wired and/or wireless communication to one or more remote devices 101, such remote devices hosting digital content data, for example text data, video content data, image data , etc. According to a variant, the data or part of these data are stored on one or more databases 102 connected or associated with one or more remote devices, for example via a wired connection of the Ethernet and/or optical fiber type.

The communication infrastructure connecting the device 10, the remote device(s) 101 and/or the database(s) 102 corresponds to one of the following infrastructures or to an infrastructure combining several of the following infrastructures:
- infrastructure of a wireless communication network, for example a 4G or 5G type wireless cellular network or a WLAN type network (from the English “Wireless Local Area Network” or in French “Réseau local sans fil”) ;
- infrastructure of a wired communication network of the WAN type (from the English “Wide Area Network” or in French “Réseau extendée”) or LAN (from the English “Local Area Network” or in French “Réseau local”) ;
- infrastructure of a fiber optic network.

The remote device(s) correspond for example to servers, for example to one or more servers of an automobile manufacturer and/or one or more servers or computers of dealers of the automobile manufacturer and/or one or more servers or computers of service providers. Car manufacturer. The remote device(s) 101 and databases 102 belong for example to a public network 100 (internet) and/or to one or more private networks (intranet).

There schematically illustrates a process for classifying data representative of a request for repair of at least one vehicle component, according to a particular and non-limiting exemplary embodiment of the present invention.

The process of is for example implemented in the system 1, and more specifically in the data processing device 10, for example by one or more processors of this device 10, or by a server type data processing device 101, for example by one or more processors of this device 101. An example of hardware embodiment of such a device is described in relation to the below.

A request to cover the repair of at least one vehicle component corresponds to a credit request, that is to say a request for reimbursement, transmitted for example by a mechanic to a car manufacturer, following repairs carried out on a vehicle whose manufacturer's warranty is active or in progress.

The support request includes a set of information or data entered by the mechanic through one or more human-machine interfaces of a data processing device (for example a computer, a smartphone (of 'English "smartphone") or a tablet).

These data or information are for example published in the form of a document or digital file comprising textual data describing the repairs carried out on one or more components or organs of the vehicle.

The support request is then transmitted electronically to a server or computer of the automobile manufacturer (for example via email) via network 1 for example.

A vehicle component or organ corresponds to any element or system of a vehicle, for example an element of the engine, the bodywork, the steering, the passenger compartment, the electronic system of the vehicle.

According to the state of the art, such a support request is processed by a person at the manufacturer level to in particular associate a category or a class with this support request, for example to ensure particular treatment in depending on the category or to statistically analyze all the support requests received.

It happens that a classification error occurs during manual processing. For example, a support request concerning an air conditioning system type component can be classified or categorized with the class or category corresponding to battery or headlight type components.

A process for automatically classifying one or more requests received for repair of at least one component of a first vehicle will be described below. Such a process advantageously implements a machine learning method, for example by a neural network.

Automatic classification using a machine learning method avoids the problems of classification made by a person.

The process of classification or prediction of membership in a class or category among a set of classes or categories advantageously comprises two phases 20, 21, each of these phases comprising one or more operations.

The first phase 20 corresponds to a so-called learning or training phase of one or more classification models (also called prediction models) and the second phase 21 corresponds to a so-called production, classification or prediction phase based on the model(s) learned in the learning phase 20 and data feeding the model(s) learned.

The first phase 20 and the second phase 21 are for example implemented by the remote device 101.

According to a variant embodiment, the first phase is implemented by the remote device 101 (or by a non-cloud server, for example a server adapted to carry out learning) and the second phase 21 by the processing device 10 .

According to another alternative embodiment, learning is implemented in the first phase 20 and the parameters of the classification model(s) are refined in real time from the data and the results of the classification of this data obtained during the second phase 21.

Learning phase 20

The learning implemented in the first phase 20 advantageously corresponds to unsupervised learning from a set of data associated with a set of support requests associated with a set or group of second vehicles. Such a set of support requests includes, for example, a few dozen, a few hundred, thousands or tens/hundreds of thousands of credit requests with an associated class or category for each of them. The class or category associated with a support request is not provided for the learning data, the learning of the classification model(s) being based only on data included in the support request forms . Unsupervised learning, unlike supervised learning, does not require prior classification of the data, which makes it possible to avoid certain biases linked to such prior classification carried out by operators, for example by the mechanic having taken into account charges the second vehicle to make the required repairs. In unsupervised learning, the learning data used to learn the classification model(s) is not labeled, i.e. no label or information on the type. data or the group to which it belongs is associated with this training data.

In a first operation 201 of the learning phase 20, the remote device 101 collects, obtains or receives (for example from a memory, one or more databases and/or other remote devices) for each second vehicle of a set of second vehicles of second data representative of a request for repair of at least one component of each second vehicle.

In a second operation 202 of the learning phase 20, the second data is processed to be formatted according to a determined format or structure.

The second operation 202 thus corresponds for example to a data preparation step to homogenize the data of all the support requests taken into account for the learning phase 20. The data preparation comprises for example operations of method of automatic natural language processing and/or automatic translation operations of the text contained in support requests into a specific target language (for example English or French).

A method of automatic natural language processing, called NLP (from English “Natural Language Processing”), is known to those skilled in the art and is for example implemented according to an automatic learning method, for example implemented works via a neural network.

An example of automatic natural language processing is for example described in “Automatic natural language processing and interpretation: contribution to the development of a computer model of interpretive semantics”, published by Ludovic Tanguy (Computer science and language, University of Rennes 1) in 1997.

Text processing using an NLP method aims to prepare and homogenize texts, for example for subsequent processing. NLP processing, for example, identifies elements of the text corresponding to verbs, adjectives and terms that appear most frequently. NLP processing includes, for example, stemming or stemming operations, lemmatization operations, deletion of stop words (for example the words of, of, a, etc.), deletion of punctuation marks, separation of comments, etc.

At the output of the NLP processing model(s), texts prepared or cleaned for the following operations are obtained, such texts being stored in memory (for example temporarily) and associated with the translated texts provided as input to the NLP model(s).

In a third operation 203, a set of information is extracted from the second data, that is to say texts corresponding to the support requests, and another set of information is generated from usage data of second vehicles associated with support requests.

Such an operation is also called feature extraction operation which consists of extracting information or characteristics from the support requests and introducing new information or data from other information contained in the support requests by analysis text and/or extracted characteristics, for example the addition of a level or degree of use of a vehicle, the creation of intervals based on the vehicle's driving periods, etc.

In a fourth operation 204, the data or information obtained in the previous operations are coded or encoded in a determined format so that these data can be interpreted by the algorithms for learning the coefficients or parameters of the classification model (or prediction model).

The coding of the data is for example obtained according to one or other of the following methods:
- TD-IDF method (from the English “Term Frequency – Inverse Document Frequency” or in French “Fréquence determs – Fréquence de document inverse”) for textual data, such a method being used in so-called text mining methods to evaluate the importance of a term contained in a document; and or
- method called “One Hot Encoding” (or in French “encoding 1 among n”) for categorical data, this method consisting of encoding a variable with 'n' states on 'n' bits of which only one takes the value '1' , the number of the bit worth '1' being the number of the state taken by the variable, 'n' representing the number of categories (or classes).

In a fifth operation 205, the classification model is learned (or trained, or generated) according to any known unsupervised learning method, from the learning data coded in the previous operation 204, i.e. say from the data obtained or extracted from the second data in relation to the classification data associated with these second data obtained as input to the learning phase.

The classification of the training data is unsupervised, similarities between the training data being sought to form groups or classes (or “cluster” in English).

Learning the classification model uses, for example, one or more factor analysis methods and one or more data partitioning methods, such as the so-called k-means method, the isolation forest method, etc.

Classification via the k-means partition method (from the English “k-means”) is for example associated with a method called “Elbow” (or “elbow” in French) in order to group together the different requests for consideration. charge in groups or classes, for example to group support requests according to the organ or part of the vehicle targeted by these support requests. The principles of such a method associated with the “Elbow” method are known to those skilled in the art and for example described in the document entitled “Determining the appropriate cluster number using elbow method for K-means algorithm”, published by H. Humaira and R. Rasyidah in January 2018.

The “k-means” method groups data sets into k groups called “clusters” in English, k corresponding to an integer, the number of groups or classes being determined automatically during learning based on the partitioning of the data. learning. For this, each set is represented in a multidimensional space (for example in a two-dimensional space). Each group, denoted Gri, is represented in this space by a centroid Ci (corresponding for example to an average) whose initial value is a random value. For each added data set, a distance Di is calculated between a point in space representing this set and each centroid Ci. The data set is then added to the group GRi whose distance Di is minimum. The centroid Ci of the group Ci in which the data set has just been added is then updated taking into account the added data set. Alternatively, multiple clusters of data sets are calculated for different k numbers and one of these clusters is kept. For example, the preserved grouping is one that minimizes standard deviations between the data sets of the groups of this grouping and the centroid associated with these groups.

Training the classification model makes it possible to obtain a determined number of groups or classes corresponding to the outputs of the classification model, each class grouping together similar data. A particular class of the set of learned classes groups together, for example, the data from support requests relating to the same organ, part or component of a vehicle. Each class therefore relates to a different type of vehicle component. For example, a first class concerns the traction battery, a second class concerns the shock absorbers, a third class concerns the steering system of the vehicle, etc.

Learning the classification model also includes an automatic naming operation 206 of the different classes identified during data partitioning. Naming includes, for example, the determination and association of a name, also called a label, identifying each class.

The naming operation 206 includes for example an analysis of the second data, for example an analysis of the words contained in each credit request of a specific class. This analysis makes it possible to determine one or more naming criteria, for example a threshold representative of the occurrence of a word in support requests grouped in the same class. For example, a class is named "Battery" or "Brake" when 75, 80, or 85% of support requests associated with that class include the word "Battery" or "Brake" respectively. According to a variant, a class takes the name "Battery" or "Brake" when the word "Battery" or "Brake", respectively, appears at least 2 or 3 times in each request for support of the class (or in 80 or 90% of class support requests).

In a seventh optional operation 207, the learned classification model is validated from data representative of a set of repair support requests and associated classes, this data being not used for unsupervised learning and dedicated to validation, these support requests being for example associated with a set of third vehicles, for example different from the set of second vehicles associated with the second learning data.

In such an operation, the performance of the learned classification model is verified from a set of validation data not integrated into the training data and from a set of determined evaluation metrics (or measures), these metrics comprising for example the following metrics: so-called PR-Curve metric (from the English “Precision-Recall Curve” or in French “Courbe precision-rappel”) and/or specificity metric and/or precision metric and/or or recall metric, etc.

The parameters or coefficients of the learned (and optionally validated) classification model are transmitted to the device implementing the classification or production phase 21 when the device implementing the classification phase 21 is different from the device implementing the classification phase 21. learning 20.

Classification or production phase 21

In a first operation 211 of the classification phase, the device in charge of the classification phase receives or obtains first data representative of a request for repair of at least one component of a first vehicle.

These first data are for example received from a data processing device implemented at a garage or vehicle repairer.

According to one variant, these first data are received from a server 101, for example a server maintained by the automobile manufacturer.

In a second operation 212 of the classification phase 21, the first data are processed to be formatted according to a determined format or structure, for example a format or structure identical to that used in the learning phase 20.

The second operation 212 thus corresponds for example to a data preparation step to homogenize the first data, as described with regard to the second operation 202 of the learning phase.

In a third operation 213, a set of information is extracted from the first data, and another set of information is generated from usage data of the first vehicle associated with the support request represented by the first data.

Such an operation is also called feature extraction operation and is identical to that described in the third operation 203 of the learning phase 20.

In a fourth operation 214, the data or information obtained in the previous operations are coded or encoded in a determined format corresponding to that of the learning phase 20 so that this data can be interpreted by the classification model learned in the learning phase.

The coding of the data is for example obtained according to one and/or the other of the methods described with regard to the fourth operation 204 of the learning phase 20.

In a fifth operation 215, the first data is classified by feeding the learned classification model with the coded data.

In a sixth operation 216, the result of the classification 2150, that is to say one or more classes with optionally for each a prediction coefficient (between 0 and 1 for example) associated with each class identified as being the class the most probable, or the most probable, to which the support request 211 must be attached, is validated from a set of analytical rules using a set of parameters obtained from databases 1101, 1102, 1103 storing data representative of fault and/or component repair processes associated with the first vehicle.

The sources 1101, 1102, 1103 of these parameters or validation data correspond for example to:
- a base in which data reporting technical facts or reasons are stored to monitor a technical problem on a vehicle, from identification to resolution of the problem; and or
- a so-called TSB database (from the English “Technical Service Bulletin” or in French “Bulletin de service technique” which documents the recommended procedures for repairing vehicles based on the problems or breakdowns identified; and/or
- a base including fault codes, also called DTC code (from English “Diagnostic Trouble Code” or in French “Diagnostic fault code”).

When several classes are identified as a result of the classification 2150, one class among the plurality of classes is for example selected based on the result of the validation, the class identified as the most probable from the analytical rules being selected.

In a seventh operation 217, the result of the validated classification is transmitted electronically via a communication network such as network 1, for example in the form of an electronic mail or email, to one or more people or groups of identified and determined people.

The result of the classification is for example transmitted to the database storing the learning data 201 with the first data having served as input data to obtain the result of the classification.

Such transmission makes it possible to enrich the content of the learning base as support requests are classified in the classification 21 (or production) phase.

New training of the classification model temporally subsequent to the unsupervised learning 205 is for example implemented according to a semi-supervised learning method each time new data is received from the classification phase 21 or when a number new data from classification phase 21 has been received.

Such a process called relearning (or semi-supervised learning) makes it possible to improve learning by refining the parameters or coefficients of the classification model and/or the criteria of the class naming rules as it is developed. use in production, based on the first data and on the result of the classification 2150 of these first data which has been validated from one or more sources 1101, 1102, 1103.

According to a variant, the learning base is further enriched from result(s) of the classification of first data validated (and possibly corrected) by one of the recipients of the classification result during an additional validation operation 219.

There schematically illustrates a device 3 configured for the classification of data representative of a request for repair of at least one component of a vehicle, according to a particular and non-limiting embodiment of the present invention. The device 3 corresponds for example to the data processing device 10 or to the remote device 101 of the .

The device 3 is for example configured for the implementation of the operations described with regard to Figures 1 and 2 and/or the steps of the method described with regard to the . Examples of such a device 3 include, but are not limited to, a computer, a laptop, a smartphone, a tablet, a server. The elements of device 3, individually or in combination, can be integrated into a single integrated circuit, into several integrated circuits, and/or into discrete components. The device 3 can be produced in the form of electronic circuits or software (or computer) modules or even a combination of electronic circuits and software modules.

The device 3 comprises one (or more) processor(s) 30 configured to execute instructions for carrying out the steps of the method and/or for executing the instructions of the software(s) embedded in the device 3. The processor 30 may include integrated memory, an input/output interface, and various circuits known to those skilled in the art. The device 3 further comprises at least one memory 31 corresponding for example to a volatile and/or non-volatile memory and/or comprises a memory storage device which may comprise volatile and/or non-volatile memory, such as EEPROM, ROM , PROM, RAM, DRAM, SRAM, flash, magnetic or optical disk.

The computer code of the embedded software(s) comprising the instructions to be loaded and executed by the processor is for example stored on memory 31.

According to a particular and non-limiting embodiment, the device 3 comprises a block 32 of interface elements for communicating with external devices, for example one or more remote servers. The interface elements of block 32 include one or more of the following interfaces:
- RF radio frequency interface, for example of the Wi-Fi® type (according to IEEE 802.11), for example in the 2.4 or 5 GHz frequency bands, or of the Bluetooth® type (according to IEEE 802.15.1), in the band frequency at 2.4 GHz, or “Sigfox”® type using UBN radio technology (from English Ultra Narrow Band, in French ultra narrow band), or LoRa in the 868 MHz frequency band, LTE (from English “Long-Term Evolution” or in French “Evolution à long term”), LTE-Advanced (or in French LTE-advanced);
- USB interface (from the English “Universal Serial Bus” or “Bus Universel en Série” in French);
- HDMI interface (from the English “High Definition Multimedia Interface”, or “Interface Multimedia Haute Definition” in French).

According to another particular and non-limiting example of embodiment, the device 3 comprises a communication interface 33 which makes it possible to establish communication with other devices (such as other computers or servers) via a communication channel 330. The communication interface 33 corresponds for example to a transmitter configured to transmit and receive information and/or data via the communication channel 330. The communication interface 33 corresponds for example to a wired network of the Ethernet type (standardized by the ISO/IEC 802-3 standard).

According to a particular and non-limiting embodiment, the device 3 can provide output signals to one or more external devices, such as a display screen 340, touch or not, one or more speakers 350 and/or other peripherals 360 (for example a haptic actuator) via output interfaces 34, 35 and 36 respectively. According to a variant, one or the other of the external devices is integrated into the device 3.

There illustrates a flowchart of the different stages of a method of classifying first data representative of a request for repair of at least one component of a vehicle, according to a particular and non-limiting embodiment of the present invention . The method is for example implemented by a processing device 10 or 101 of a communication system or by the device 3 of the .

In a first step 41, the first data are received.

In a second step 42, the first data is processed to be formatted according to a determined structure.

In a third step 43, a first set of determined information is extracted from the first formatted data and a second set of information is generated from vehicle usage data.

In a fourth step 44, the first and second information are encoded in a determined format.

In a fifth step 45, the first data is classified by feeding a classification model with the first and second coded information, the classification model having been learned in an unsupervised learning phase of a machine learning method.

In a sixth step 46, the result of the classification is validated from a set of analytical rules using a set of parameters obtained from databases storing data representative of faults and/or component repair processes associated with the vehicle .

In a seventh step 47, the validated result of the classification is transmitted electronically to one or more recipients.

According to a variant, the variants and examples of the operations described in relation to the or 2 apply to the steps of the process of .

Of course, the present invention is not limited to the exemplary embodiments described above but extends to an unsupervised learning method of a data classification model of requests for support for repair of elements of vehicles which would include secondary steps without thereby departing from the scope of the present invention. The same would apply to a device configured to implement such a process.

Claims (10)

Method for classifying first data representative of a request for repair of at least one component of a first vehicle, said method being implemented by at least one processor, said method comprising the following steps:
- reception (41; 211) of said first data;
- processing (42; 212) of said first data to format said first data according to a determined structure;
- extraction (43; 213) of a first set of determined information from said first formatted data and generation of a second set of information from usage data of said first vehicle;
- coding (44; 214) of said first and second information in a determined format;
- classification (45; 215) of said first data by feeding a classification model with said first and second coded information, said classification model having been learned in an unsupervised learning phase (20) of a machine learning method ;
- validation (46; 216) of a result of said classification from a set of analytical rules using a set of parameters obtained from databases (2101, 2102, 2103) storing data representative of fault and/or component repair process associated with said first vehicle;
- electronic transmission (47; 217) of the validated result of said classification (45; 215). Method according to claim 1, for which said learning phase (20) comprises the following steps:
- obtaining (201), for each second vehicle of a set of second vehicles, second data representative of a request for repair of at least one component of said each second vehicle;
- processing (202) of said second data to format said second data according to said determined structure;
- extraction (203) of a third set of determined information from said second formatted data and generation of a fourth set of information from usage data of said each second vehicle;
- coding (204) of said third and fourth information in a determined format;
- unsupervised learning (205) of said classification model from said third and fourth information coded in relation to said second data. A method according to claim 2, wherein said unsupervised learning comprises determining a plurality of classes and determining (206) a name for each class of said plurality of classes from said third and fourth information encoded in relation to said second data. Method according to claim 3, for which said unsupervised learning further comprises determining a set of naming criteria for said plurality of classes from said third and fourth information encoded in relation to said second data. Method according to one of claims 2 to 4, for which said learning phase (20) further comprises a step of validating (207) said classification model from third data representative of a set of processing requests. repair charge for at least one component of each third vehicle of a set of third vehicles, fifth data representative of class of said determined set of classes being associated with said third data. Method according to one of claims 2 to 5, for which said learning phase further comprises a step of semi-supervised learning of said classification model subsequent to said unsupervised learning, said semi-supervised learning being obtained as a function of said first and second encoded information associated with a validated classification result obtained from said validation. Method according to one of claims 1 to 6, for which said processing of said first data comprises processing by natural language processing method. Method according to one of claims 1 to 7, for which said at least one processor implements a neural network. Computer program comprising instructions for implementing the method according to any one of the preceding claims, when these instructions are executed by a processor. Device (3) for classifying first data representative of a request for repair of at least one vehicle component, the device (3) comprising a memory (31) associated with at least one processor (30) configured for implementing the steps of the method according to any one of claims 1 to 8.