Recommendation of textual data in the process of acquisition

Abstract

An information-processing device for providing assistance with acquisition of required textual data (S13) is proposed, the method comprising: - establishing a first semantic graph (S5) based at least on said required textual data, - obtaining textual data from data delivered by at least one source (S2), and establishing at least a second semantic graph (S4) based on the obtained textual data, - performing a search (S6) for at least partial similarity between the first and second semantic graphs in order to identify at least some of the required textual data among the obtained textual data (S7), and - generating a recommendation based on the identified textual data.

Description

Recommendation of text data being acquired

[0001] This disclosure relates to the field of data processing by multiple computer applications.

[0002] In everyday life, particularly in the professional world and particularly in the context of the "digital enterprise", users require the execution of various IT processes. In particular, they may contribute to multi-application or multi-media projects (the execution of which may require several applications (or media)).

[0003] For example, a user may open an application, search for information useful for a current task, then open a second application and enter text data by copying some of the information previously found.

[0004] In another example where a user is reading documentation or a message from a conversation, they can trigger a parallel action in another business application to enter text data related to the documentation or message being viewed (or vice versa).

[0005] Managing these multiple applications by a user can be tedious and a computer-based way is sought to facilitate navigation between these applications. Summary

[0006] This disclosure improves the situation.

[0007] To this end, it proposes a process implemented by a computer device, comprising:

[0008] - obtaining textual data from data from at least one source (S2), and - issuing a recommendation relating to data required by at least one computer application currently running and with which at least one user interface of said device allows interaction, based on a semantic similarity between at least some of said obtained data and at least some of said required data.

[0009] In at least one embodiment, the method may comprise identifying at least a portion of the required textual data from among the obtained textual data (S7), and said recommendation is based on the identified textual data.

[0010] In at least one embodiment, said required data is data being written by a user of said device.

[0011] In at least one embodiment, said recommendation comprises a proposal of elements correcting and / or supplementing said required data.

[0012] In at least one embodiment, said semantic similarity takes into account a similarity between a first semantic graph established for said obtained data and a second semantic graph established for said required data.

[0013] Thus, in at least one embodiment, the method comprises: - establishing a first semantic graph at least as a function of required textual data, - obtaining textual data from data from at least one source, and establishing at least a second semantic graph as a function of the textual data obtained, - carrying out an at least partial similarity search between the first and second semantic graphs to identify at least part of the required textual data among the textual data obtained, and - issuing a recommendation as a function of the identified textual data.

[0014] Here, the term "establishing a semantic graph" means a semantic analysis (and not simply lexical) making it possible to structure the textual data according to, for example, a tree structure. Thus, the aforementioned similarity search may include the identification of one or more branches comprising common elements (nodes for example) between the first and second graphs. The establishment of such graphs typically makes it possible to establish a structure of the textual data required to easily identify the textual data from the source which may correspond to this required data.

[0015] The data from the source is deemed reliable and can thus constitute reference data. Thus, it is possible to supplement and / or correct data from the first graph based on the data from the second graph. The method then provides assistance to a user to easily enter reliable text data, which may be required for example by an application in embodiments presented later.

[0016] In at least one embodiment, the method may then comprise: - establishing the first semantic graph at least as a function of textual data to be verified, - carrying out an at least partial similarity search between the first and second semantic graphs to identify at least part of the textual data to be verified among the textual data obtained, and - issuing a correction recommendation if data from the identified part differs from the corresponding textual data obtained.

[0017] Such an embodiment can thus, at least in certain embodiments, help to detect for example errors in required data and typically being entered, and then recommend corrections of these errors.

[0018] As indicated above, the required data may be required, for example, by a computer application that is running and requesting the required data. In at least one embodiment, the method may then further comprise cooperation with such a computer application. For example, such an application may animate the presentation of an online form on an active web page, and thus wait for (i.e., request) data to be entered by a user. According to another example, such an application may convert image data currently being acquired by a camera into text data (by optical character recognition). These images may represent, for example, characters on a sheet of paper (e.g., a paper form) or a whiteboard, in particular handwritten characters, currently being entered by a user.For example, in this case, it may be advantageous to implement a verification of the text data already entered by the user to suggest corrections, or even recommend new data to enter.

[0019] To generate these recommendations, the method may comprise, in one embodiment: - animating a human-machine interface connected to the computing device, and - sending a message via the human-machine interface corresponding to said recommendation.

[0020] This message may lead to the suggestion of data to be entered, to propose to the user the textual data to be provided to the aforementioned computer application. An example of such an implementation is illustrated on theshowing a window which opens (in "pop-up") to suggest help while the user attempts to enter data, and if he accepts the proposed help, the online form is filled with the required data.

[0021] According to the embodiments, the human-machine interface may be an interface integrated into the device or an interface coupled to the device via wired and / or wireless communication means.

[0022] Thus, in exemplary embodiments, the human-machine interface may comprise a screen displaying at least a first window specific to the computer application, and the method comprises: - controlling a display on the screen of said recommendation in a second window, simultaneously with the first window.

[0023] Such an implementation may, at least in certain embodiments, help to improve the ergonomics of the user interface, the recommendation being displayed for example “above” a window specific to the current application (for example transparently).

[0024] In some embodiments, alternatively or in addition, the aforementioned recommendation may directly include the identified textual data, and the method comprises:- feeding the computer application with the identified textual data. Thus, for example, the textual data required by the application is directly provided to the application, without user intervention. In the example illustrated in the, the data for a destination city “Rennes” and travel date “January 7, 2022” are directly filled in the online form. The user can of course correct this data later, if necessary.

[0025] In this example, the required data appears in an image shown on the which corresponds to an instant messaging chat for example. This image is considered here as a reference data source and the online form to be completed (accessible via a web page for example) of the corresponds to the application requesting the required data.

[0026] A semantic analysis of this required data can lead to the first graph mentioned above and a semantic analysis of the cat can lead to the second graph mentioned above.

[0027] In at least one embodiment, these first and second semantic graphs are tree-like and comprise: * nodes representing predicates, and * leaves representing textual data and responding to the predicates. The method can then comprise: - identifying nodes with common predicates between the first and second graphs, and - selecting in the second graph leaves from the nodes with common predicates, said selected leaves corresponding to said part of the textual data required from among the textual data obtained.

[0028] Thus, in such an implementation, a search for common nodes in the graphs is carried out to deduce the leaves from these nodes in the second graph and which correspond to relevant responses to provide the required data.

[0029] In the example of Figures 4 and 5, it is possible to identify the common nodes under the d / date-entity branch corresponding to a future destination date. This date of the second graph of the giving the answer to the data required under the same nodes of the first graph of the. Thus, the date (for example by default) of February 10, 2022 of the first graph must be corrected to the date of January 7 given by the second graph.

[0030] For example, these first and second semantic graphs may be in the computer language of abstract semantic representation, or "AMR" (for "Abstract Meaning Representation").

[0031] The data from the sources can be in different forms. For example, it can be image data containing text, and the method can then include:- processing this image data by character recognition to obtain text data.

[0032] Thus, the reference text data can come from an image such as a digital photograph, a digital video, in any format.

[0033] In some embodiments, additionally or alternatively, data from a source may be sound data, and the method may comprise:- processing said sound data to detect speech signals and converting the speech signals into text data.

[0034] In this case, the reference text data comes, for example, from an audio recording.

[0035] In some embodiments, additionally or alternatively, data from a source may be from a handwritten character input computing device, and the method may include:- implementing handwritten character recognition to convert the input characters into text data.

[0036] For example, the reference text data here could come from a connected whiteboard or a graphics tablet with handwriting recognition, or others.

[0037] In certain embodiments, alternatively or in addition, the required textual data may itself come from an analysis of image data of the aforementioned type, or from audio content, or from a file acquired by handwritten character recognition.

[0038] Of course, the reference data and / or the required data may come from editable text files (for example in “doc” or “docx” format) or from non-editable text (for example in “pdf” format). In one embodiment, at least part of the textual data obtained may be recorded in the memory of the computing device with data relating to a type of source from which they come. Thus, for example, in the event of a conflict between the respective responses given by different sources, certain reference data (non-editable for example) may “take precedence” over others (for example, editable text which may typically contain input errors). According to another example, reference data coming from a trusted source, such as an official source (for example, an institutional or government site) may take precedence over other reference data coming from any source.

[0039] In one embodiment, the data from at least one source are recorded in the memory of the computing device for a first duration (which can be chosen and for example configurable), after a final use for the implementation of the method.

[0040] Such an implementation makes it possible to keep in memory (buffer for example) the most frequently consulted data and to erase on the other hand reference data which has not been used since the first duration, mentioned above.

[0041] In one embodiment, the method may comprise implementing an artificial intelligence programmed to learn, based on user feedback, relevant recommendations, and select after learning a relevant recommendation to issue based on the identified textual data.

[0042] For example, if multiple graph sheets are possible when there are more than two graphs to compare, artificial intelligence can help to immediately discard sheets corresponding to recommendations that would be rejected by the user.

[0043] According to another aspect, there is provided a computer program comprising instructions for implementing all or part of a method as defined herein when this program is executed by a processor. According to another aspect, there is provided a non-transitory, computer-readable recording medium on which such a program is recorded.

[0044] According to another aspect, there is provided a computing device comprising a processing circuit configured to implement the method according to the present invention.

[0045] Other features, details and advantages will become apparent upon reading the detailed description below, and upon analyzing the attached drawings, in which: Fig. 1

[0046] illustrates an example of a method according to a possible embodiment. Fig. 2

[0047] illustrates an example of a reference data source for establishing a “second graph” as defined above. Fig. 3

[0048] illustrates an example of a data acquisition context, here an application running and requesting text data. Fig. 4

[0049] illustrates a “second graph”, established according to the source illustrated on the. Fig. 5

[0050] illustrates a “first graph” (as defined above), established from the data required in the context of the. Fig. 6

[0051] illustrates an interface animation to offer a recommendation to a user. Fig. 7

[0052] illustrates an example of a computing device for implementing the method.

[0053] La illustrates, by way of example and in a non-limiting manner, an implementation of the method which is the subject of the present application by a computer device. With reference to the, during a first step S1, a context of use is detected. This context can be detected for example if an application requiring textual data is currently running (test of step S11) and for example if the user of the device has recently consulted or is still consulting a document (test of step S12). This document can be text in a non-editable format (type "pdf" for example), or a digital image, or text in an editable format, or others. Here, "recently" means the fact that the consultation of this document dates back less than an hour for example, or even less than a few days, depending on the relevance of the document (as presented later with reference to steps S8 and S9).

[0054] In this example, a user may enter text data as part of the execution of the application in step S11. For example, this application may formulate questions via a human-machine interface of the device, and user responses are expected to be entered. For example, this may be an online form to be completed by the user, as discussed later with reference to the. Thus, text data (TXT1) is required by the application in step S13.

[0055] Of course, the determination of this context according to steps S11 and S12 is an optional example and represented for this purpose by dotted lines. In a variant, the user can for example write information on a whiteboard or a white sheet, and a camera (for example on connected glasses) acquires the information during writing. This context can also be identified within the meaning of step S1 and recognition of characters written on the sheet or the table can be carried out to:- determine textual data during writing by the user, and- identify whether such data can be corrected (for example a form filled out by hand by a user as in an exemplary embodiment presented later) or simply completed.

[0056] Thus, in such an embodiment again, text data are required within the meaning of step S13, in order to be able to carry out a verification and possibly a correction of the data being entered or to complete the data being entered. This verification and / or completeness of the data being entered can be carried out on the basis of reference text data, as explained below.

[0057] Referring now to step S2 of the, reference text data TXT2 comes from one or more sources (other than the application currently running for example). For example, these sources can be files stored at least temporarily in a memory MEM of a device according to the present description, or web pages recently (or frequently) consulted by the user. For example, files (image files comprising text, or editable text files, or others) can be stored in a dedicated memory area, to be consulted by the method described here, in order to identify text data TXT2 which can be used as reference data. In the case of an image file for example, character recognition can participate in determining whether the text data resulting from the conversion of the image to the text can be used as reference data.Optionally, these files (or links to these files) may be stored in the aforementioned memory area with an indexing of the nature of the data that may be used, for example: - data specific to the user's identity (data from a user's identity card, or social security or other), - data acquired in a professional context, - etc.

[0058] Referring now to step S14 of the, a semantic analysis of the questions asked in the execution of the application, or of the textual data being written by the user, makes it possible to establish a semantic graph in step S5. This semantic graph GRAPH1 can be for example in the form of a tree in abstract semantic representation, for example in “AMR” (for “Abstract Meaning Representation”). It is recalled that AMR is a computer language for semantic representation. AMR graphs can represent one or more entire sentences. AMR differs from a syntactic analysis in particular in that different sentences, but with similar meaning, can have the same AMR graph, even if they are not formulated in an identical manner.A tree graph, particularly of the AMR type, then has nodes and leaves, the leaves being answers to questions represented by nodes (or "predicates") upstream. For example, "5:30 p.m." can be a leaf of a predicate node "what time?".

[0059] Thus, in the detailed embodiments, the establishment of this graph GRAPH1 can make it possible to establish a structure of the textual data required to identify the textual data of the source which can correspond to the required data. Indeed, following the establishment of this graph GRAPH1 in step S5 of the, the method also proposes a semantic analysis of the reference data TXT2 in step S3, as well as the establishment of a semantic graph GRAPH2 of the same nature (for example AMR) in step S4, based on these reference data. The selection of a source in step S2 can also be a function of the textual data required if the files corresponding to the different sources are indexed in memory according to particular themes.For example, files related to the topic "user identity" can be selected and analyzed semantically (after possible conversion into textual data), if predicates of the graph GRAPH1 of the required data TXT1 correspond to this topic (by asking the question "social security number?" for example).

[0060] Step S6 then consists of comparing the two graphs GRAPH1 and GRAPH2 to identify similarities, particularly at the predicate level. For example, if common nodes are identified between the two graphs, the leaves of the second graph GRAPH2 originating from these nodes may be answers to questions formulated in the predicates of the first graph GRAPH1. Thus, these answers may correspond to the textual data required from step S13. At the end of this step S6, the reference textual data TXT2 may thus make it possible to formulate recommendations in step S7, aimed at answering the questions of the first graph GRAPH1, or possibly at correcting leaves of the first graph GRAPH1. In at least some embodiments, these recommendations, in the form of textual data, may directly feed the application being executed (without user intervention), or even be presented to the user in step S15.

[0061] At this step S15, for example in an application where the user fills out a form (online on a web page, or by hand on a sheet of paper), a human-machine interface can be animated to display on a screen a window presenting a text corresponding to the recommendation (in transparency or next to the form of the web page, or on the lenses of connected glasses in augmented reality when the user writes on a sheet or a table). Alternatively, or in addition, the human-machine interface can include an audio headset and a recommendation message can be played on this audio headset (the message indicating for example the required text data or a recommendation to correct text entered by the user).

[0062] Thus, the reference data from the aforementioned sources in step S2 make it possible to assist in providing data required by an application or by the user in an activity of entering such data (this assistance being able to be provided via the animation of the aforementioned human-machine interface).

[0063] The sources providing these reference data may, for this purpose, be stored in memory as indicated previously. However, to avoid loading the memory and / or to avoid proposing irrelevant reference data, a test may be conducted in step S8 to determine whether the user validates the recommendation by entering the required data. If this is not the case, the references of the file as a source of reference data may be deleted from the memory in step S10. For example, an artificial intelligence may be implemented to determine the data sources to be retained and which may be relevant, by learning according to the validations of the recommendations made by the user. In addition or as a variant, in step S9, if a source has not been used for a threshold duration (for example a few days or more), the reference to this source in the memory may be deleted in step S10.

[0064] Thus, such an implementation can in particular help the user to enter data, by directly feeding the application requiring this data (the user being able to check, for example, that the data in an online form, filled in directly without his intervention, are correct), or by proposing an animation of a human-machine interface. It is thus proposed an improvement (for example an optimization) of the interactions between applications that a user can use by positioning himself, for example, "above" (along a Z axis "of a display screen" as illustrated in the figure below), or next to it (along X, Y axes) or more generally via the superposition of information in the case of augmented reality glasses, by combining and linking the information useful to the user's activity.

[0065] Actions between applications are assumed to be interconnected, for example according to preferential rules that the user sets in the aforementioned computer program (and which can be refined by learning artificial intelligence). The application of these rules may follow, for example, the detection of words or sentences visible on the screen (independently of the application currently in use). Thus, computer modules such as an optical character recognition (or "OCR") application, an abstract semantic representation (or "AMR") computer application, a computer program for identifying similar predicates in graphs and searching for answer sheets may be implemented.

[0066] Examples of techniques for identifying an acquisition context within the meaning of step S1 are presented below, with the aim of proposing, for example, a human-machine interface animation in a context of restitution of the required data.

[0067] A multi-source assistance module may be provided requiring inputs from a context analyzer module. A first functional building block identifies, for example, on the user's screen(s) the applications currently in use. If it is an implementation in an augmented reality environment, the first functional building block identifies the real or virtual areas that have the user's attention. This identification may be carried out by eye tracking, and / or by tracking presses on a touch surface (or "pad"), and / or by tracking another input interface such as a mouse (or "mouse tracking"), or more generally by tracking system events.

[0068] This step allows, for example, to extract images from the application areas observed by the user or to produce such images.

[0069] For a user who observes two distinct applications regularly (for example with successive focus towards the two applications and / or an observed shift of gaze, detected by matching with X,Y positions of the applications present on the screen) and for a configurable period of time, two images are therefore copied and associated with this period of time T (according for example to the S12 test presented above).

[0070] This illustrates an example of an application in which a user of an instant messaging application tells a correspondent that he or she wants to go to a city on a certain date. At the same time, he or she consults another online transport booking application to get to this city according to the. Two images are then generated corresponding to Figures 2 and 3.

[0071] Of course, the scope of the method can be extended to a virtual reality or augmented reality context. For example, augmented reality glasses can observe and represent a (real) whiteboard and implement a process for recognizing handwritten characters on this board from the acquisition of a real image of this board. A television screen that reproduces information can also be the subject of the acquisition of an image. Thus, it is possible to carry out a service composition in a mixed reality world (virtual / real).

[0072] Optical recognition of the characters present on the images thus copied can be carried out, as well as a precise analysis of the characteristics of these text zones (to identify an editable zone or a label, for example). The recovery of the characters can be carried out for example via a request intended for the operating system of the computer equipment used, or, to locate an editing zone, by an analysis of the appearance of a mouse pointer when hovering over the zone containing the characters (the pointer of a mouse is no longer a simple arrow on an editable zone for example), or by the appearance of the zone itself (for example with or without a border, a different background color, or others).

[0073] The example in Figures 2 and 3 concerns solid text fields, but it is also possible to access text content from more complex components such as a drop-down list, a radio button with different headings written depending on the position of the button, or others.

[0074] The apparent texts can then be extracted from the images. Their positions in the images can be stored in memory. A semantic analysis of the apparent texts is carried out to establish a semantic graph for each image.

[0075] For example, in the cases of figures 2 and 3, in image 1 corresponding to: "I would need to go to Rennes on Monday, January 7th", the text attributes can be "position I1: X,Y / non-editable text (= NE)".

[0076] In image 2 (), the text attributes are:- “my travel management app” at position I2: X,Y (NE),- “flight / train” “hotel” “car” “door to door” at position I3: X,Y ,- “round trip” “one way” at position I4: X,Y ,- “office” at position I5: X,Y- “To enter a location” at position I6: X,Y- “When?” at position I7: X,Y (NE)- “Go” at position I8: X,Y (NE)- “Thursday, February 10, 2022” at position I9: X,Y- “12:30 p.m.” at position I10: X,Y ,

[0077] etc.

[0078] A semantic graphical representation of the extracted texts can, for example, be achieved using the Abstract Semantic Representation (AMR) technique. Other techniques for graphically rendering the meaning of sentences can also be used.

[0079] The semantic graph presents topics and arguments according to a tree as illustrated as an example in the. The tree of the corresponds more precisely to the semantic representation of image 1 of the. For at least one node of the tree (for example for each node), it can also be stored in memory the attributes such as the X / Y position of the text and its characteristics (editable or non-editable text).

[0080] The text extracted from image 2 could be: "my travel management application" "flight / train hotel car" "door to door" "round trip / one way" "from office" "position A" "enter a location" "When? Go" "Thursday, February 10, 2022" (this last indication coming by default for example).

[0081] In some contexts, the graphs produced may be very similar, and graph comparison alone may reveal correspondences between data from images 1 and 2 displayed on the screen. However, in some cases, the predicates may be different but still semantically close. For example: "t / travel" and "g / go", with respect to their arguments, are similar ("start point", "end point", "traveller", "goer") (where the argument "goer" represents an entity in motion).

[0082] However, in some embodiments, pre-programming may be provided listing general similarities between predicates so that AMR predicate similarity can be identified between two graphs to perform a correspondence between information of these respective graphs.

[0083] Thus, a GRAPH1 graph corresponding for example to an editing zone on the screen of the device (for example the form of the) can thus be completed (at least partially) by elements of one or more other GRAPH2 graphs (elements of "inputs") of data from other sources (for example the recent chat of the). Thus, these elements can be proposed in a recommendation. This recommendation can be displayed for example in the form of a window which opens on the user's screen (animation of the human-machine interface of step S15 above), "above" the other windows as illustrated in the in order to propose these elements to the user and, if the latter accepts it, to fill, with these elements, the data required in the current application ("Rennes", "Monday January 7, 2022"). The text proposed in semantic context can be directly entered in the application if the user agrees with the recommendation made (i.e.if the device's understanding of the task to be performed seems correct to it).

[0084] The user interface window that opens can, for example, be positioned on top of other active windows and as close as possible to the applications concerned, for example transparently, as illustrated in the figure. It can also be animated (alternatively or in addition) with a sound interface (with voice synthesis for example) to guide the user in entering numbers for example or to spell a city, or others.

[0085] Recommendation generation can also implement a learning module. For example, if the action proposed in the created interface is executed correctly (and therefore validated by the user), it can then be resubmitted in future uses. A recommendation window can also be implemented when the semantic links have been validated during previous uses, typically.

[0086] In another use case below, corrections are recommended to the user, following the entry of text data, carried out by the user.

[0087] In this scenario, here of mixed reality, the user has augmented reality glasses. Such glasses are equipped with a camera module capable of acquiring an image that the user views and analyzing this image in particular to detect typically written characters. The user views, for example, a social security card for a few seconds. The above method can be implemented to convert the acquired image data into textual data (by character recognition). The following information is thus extracted, for example: - "health insurance card: 290042216205897", and - "issued on August 9, 2002".

[0088] The GRAPH2 semantic graph of this reference data makes it possible to identify sheet 290042216205897 as corresponding to a predicate of type “n / number”.

[0089] The user then fills out an online form that asks: "Please enter your health insurance card number: XXXXXX." However, in doing so, the user mistakenly enters: "290042216205889."

[0090] The semantic analysis of the question posed establishes a graph GRAPH1 in which the expected answer follows the predicate of type "n / number". However, the answer entered by the user does not correspond to the answer of the reference graph GRAPH2 which follows the predicate of the same type in a similar branch.

[0091] An error in the entered number is thus detected and a pop-up window warns the user through a graphical interface (with a superposition of the recommendation message as close as possible to the erroneous information). Alternatively, or in addition, an audio interface can emit an error message (for example a beep or a voice message giving the recommendation: here, the valid number) on an audio headset, according to the user's preferences.

[0092] The above method can thus be applied in different environments, in particular entirely digital work environments and / or in virtual or augmented reality, and offer, in at least one embodiment, assistance in the use of at least one application. For example, the method can help users to combine different applications to complete a task. In particular, the implementation of the method, as illustrated in the use cases above, can be carried out in real time, and provide a dynamic, contextual aspect of the managed data (as opposed to “cold data”).Furthermore, at least in certain embodiments, the sources used (step S2) can provide a local character to the execution of the method (for example without using internet data, if necessary), thus offering advantages in terms of data protection (or “privacy”) unlike using a shared knowledge base for example, and this from a user experience in progress on a device such as a terminal, typically.

[0093] An example embodiment of such a device DIS has been illustrated, for implementing the method defined above, which may typically comprise a processing circuit equipped with at least: - a first interface INT1, for receiving reference data (for example in the form of image files, or data from a connected table, or others) from different sources (SOU1, SOU2, etc.), this data being able to be recorded in memory MEM, at least temporarily, - a processor PROC and the memory MEM, this memory MEM being able to further store in particular instruction code data of a computer program within the meaning of the present disclosure, the memory being accessible by the processor PROC to read these instruction codes and execute the method presented above, and - a second interface INT2 controlling equipment of a human-machine interface HMI such as a screen (for opening recommendation windows) and / or an audio headset, or others.

Claims

A method implemented by a computer device, comprising: - obtaining textual data from data from at least one source (S2), and - issuing a recommendation relating to data required by at least one computer application in execution and with which at least one user interface of said device allows interaction, based on a semantic similarity between at least some of said data obtained and at least some of said data required. A method according to claim 1 comprising an identification of at least a part of the required textual data among the obtained textual data (S7), and wherein said recommendation is a function of the identified textual data. A method according to claim 1 or 2 wherein said required data are data being written by a user of said device. A method according to claims 1 to 3, wherein said recommendation includes a proposal of elements correcting and / or supplementing said required data A method according to any one of claims 1 to 4 wherein said semantic similarity takes into account a similarity between a first semantic graph established for said obtained data and a second semantic graph established for said required data. The method according to claim 1 to 5, the method comprising: - establishing the first semantic graph at least as a function of textual data to be verified, - performing at least a partial similarity search between the first and second semantic graphs to identify at least a part of the textual data to be verified among the textual data obtained, and - issuing a correction recommendation if data from the identified part differ from the corresponding textual data obtained. A method according to any one of claims 1 to 6, comprising: - animating a human-machine interface connected to the computer device (S15), and - sending a message via the human-machine interface corresponding to said recommendation. Method according to claim 7, wherein the human-machine interface comprises a screen (HMI) displaying at least one first window specific to the computer application, and the method comprises:- controlling a display on the screen of said recommendation in a second window, simultaneously with the first window (FIG.6). A method according to any one of claims 1 to 8, wherein the recommendation includes the identified textual data, and the method comprises: - feeding the computer application with the identified textual data. A method according to at least one of the preceding claims, wherein the first and second semantic graphs are tree-like and comprise: * nodes representing predicates, and * leaves representing textual data and corresponding to the predicates, the method comprising (FIG.4, FIG.5): - identifying nodes with common predicates between the first and second graphs, and - selecting in the second graph leaves from the nodes with common predicates, said selected leaves corresponding to said part of the required textual data among the obtained textual data. A method according to at least one of the preceding claims, wherein said first and second semantic graphs are in abstract semantic representation computer language, or "AMR". A method according to at least one of the preceding claims, wherein said data from at least one source are image data containing text, and the method comprises: - processing said image data by character recognition (S2) to obtain textual data (TXT2). A method according to at least one of the preceding claims, wherein said data from at least one source are sound data, and the method comprises: - processing said sound data to detect speech signals and converting the speech signals into text data. A method according to at least one of the preceding claims, wherein said data from at least one source are data from a computer device for inputting handwritten characters, and the method comprises: - implementing handwritten character recognition to convert the entered characters into textual data. A method according to at least one of the preceding claims, wherein said data from at least one source are recorded in the memory of the computer device for a first duration (S9), after a final use for the implementation of the method. A method according to at least one of the preceding claims, comprising the implementation of an artificial intelligence programmed to learn, based on user feedback, relevant recommendations (S8, S10), and to select after learning a relevant recommendation to issue based on the identified textual data.

Images