Communication method and system for determining a sequence of services associated with a conversation

The present invention relates to the management of information exchanged during a conversation, and more particularly to communication methods and systems for identifying the order of services associated with a conversation in which a user participates from a communication system.

During a telephone conversation or a conversation that takes place through instant messaging software (also known as "chat"), the interlocutor often assigns work related to the subject the interlocutor is discussing.

For example, if the conversation between two interlocutors concerns the preparation of an event, such as an excursion to a tourist destination, the interlocutors may assign during the conversation work related to the planning of the excursion. .. Therefore, after the conversation, one of the interlocutors can look up the weather forecast for the place, find a route to the place, book a train ticket, or contact a third party. It is possible to be in charge of asking a third party if they are interested in such an excursion.

Traditionally, to look up the information requested during the conversation or to perform the assigned task, the interlocutor uses a computer application, such as a web browser, installed in the interlocutor's communication system. , Activate the service.

In the example above, the interlocutor responsible for investigating the weather opens a web browser installed in the interlocutor's communication system and enters the URL (uniform resource locator) corresponding to the weather forecast website. can do. In addition, if the interlocutor also has to book a transportation ticket, the interlocutor can enter a new URL in the web browser that corresponds to the website for booking the ticket online. is there. Similarly, if, during the conversation, the interlocutor offers or is asked to send an e-mail to the friend inviting the friend to participate in the excursion, the interlocutor sends the e-mail. It is possible to activate the e-mail service available from the interlocutor's communication system so that it can be transmitted.

However, the interlocutor must not forget the work assigned to the interlocutor during the conversation. The interlocutor does not yet know the services available from the interlocutor's communication system, or performs such services, which allows the interlocutor to perform the interlocutor's work assigned during the conversation. I haven't thought about it yet.

In addition, if the interlocutor identifies suitable services, in order to launch those services, the interlocutor opens a particular application from the interlocutor's communication system and obtains the information already mentioned during the conversation. Must be entered at work. This task is even more cumbersome because it must be performed each time the interlocutor needs a new service to perform a task.

<p> An object of the present invention is to facilitate the execution of tasks assigned to a user during a conversation.</p>

<p> With the above in mind, the first object of the present invention is a method of communication for specifying the order of services associated with a conversation in which a user participates from a communication system, and when the conversation is completed, With the steps to recover all the expressions used during the conversation, Steps to semantically analyze the recovered expression, Select multiple services from a list of pre-registered services, at least according to their semantic analysis, and each service in this list is a step that can be executed from that communication system. Steps to contextually analyze the expressions used during the conversation, Depending on this context analysis, the steps to determine the order of those selected services, The present invention relates to a method comprising a step of displaying the order of selected services on the communication system.</p><p> A second object of the present invention is a communication system including means for enabling a user to have a conversation. At least a means of storing a list of services that can be launched from that communication system, Select multiple services from the list of stored services based on at least one semantic analysis of all expressions used during the conversation, and contextual analysis of all expressions used during the conversation. With a natural language analysis server that can determine the order of selected services based on The present invention relates to a system further comprising means for displaying the order of selected services.</p><p> Other features and advantages of the present invention will become apparent by reading the following description with reference to the accompanying drawings.</p>

<figref num="1">FIG. 5 is a schematic diagram showing a preferred architecture of various elements constituting a communication system for specifying the order of services according to the present invention, and a dialogue between the elements.</figref><figref num="2">FIG. 6 illustrates various interactions between communication system users, messaging clients, and natural language analysis servers according to preferred embodiments of the present invention.</figref><figref num="3">It is a figure which shows the graphic interface of the instant messaging client which displays the order of services according to this invention.</figref>

Communication system 1 shown in FIG. 1 is a terminal device 10, such as telephone 10 or computer 11, which is connected to a screen and further allows user 12 to have a conversation over the telephone or through an instant messaging client. Conventionally, 11 sets are provided.

Such communication system 1 can also act as an access point to the service. So, for example, if a web browser is installed on computer 11, user 12 of communication system 1 can access the weather forecast service, query a bank account, or, among other things, a transportation ticket from this computer 11. Such services are available on internet sites accessed from this web browser.

Similarly, if an Outlook® type messaging client is installed on communication system 1, user 12 can access the messaging service and therefore receive electronic messages from other users. Or you can send an electronic message to another user.

It is also possible to envision the possibility of launching a conference call service from one of the terminal devices 10 and 11 of the communication system 1 and then executing that service.

According to the method of the present invention, when the user 12 has a conversation, at the end of the conversation, all the expressions used during the conversation can be taken out and sent to the natural language analysis server 13. This retrieval step is performed after a voice conversation over the telephone 10, or is performed, for example, via an installed instant messaging client 14 on terminal device 11 of communication system 1, satisfying the conditions. It can be performed after a written conversation.

Next, the natural language analysis server 13 executes a semantic analysis of the retrieved expression, and selects a plurality of services from a list of pre-stored services based on the executed semantic analysis. Each service in this list can consist of one of the aforementioned services that can be launched from communication system 1 and / or run.

When the service is selected, the natural language analysis server 13 performs contextual analysis of the expressions used during the conversation. Context analysis refers to analyzing an expression based on terms used before or after the expression in conversation, unlike semantic analysis, in which only the meaning of the expression is analyzed.

After this context analysis, the ordering of the selected services is determined based on the results of the context analysis. The identified order is then displayed on the interface of communication system 1.

With reference to Figures 2 and 3, the various features of the communication system, and the various stages, that are implemented to identify the order of services associated with the conversation that took place, are carried out by the instant messaging software. If so, it will be explained in detail.

As shown in FIGS. 2 and 3, user 12 is talking to at least one other user via instant messaging software. Such conversations are initialized by opening a shared session from each instant messaging client installed in the user's communication system.

In this case, for the user, the conversation is first entered through the interface of the user's instant messaging client and sent one after the other, and secondly entered by the other participants in the conversation. , Includes receiving the transmitted representation on the interface of the user's client. The expressions entered and received are traditionally words or phrases. Traditionally, one of the display areas 100 that make up the interface of an instant messaging client is intended to display all of the representations sent and received by the user during a conversation.

Figure 3 shows a portion of the messaging client's graphic interface where a user named Matthew had a conversation with a second user named Alan. All of the expressions exchanged during the conversation are displayed on the display area 100 of the interface.

According to the method of the present invention, at the end of a conversation, Matthew can trigger a feature for scheduling work from Matthew's instant messaging client (step E1 in Figure 2). All of the expressions displayed in the display area 100, which indicate all of the expressions used during the conversation, are then sent to the natural language analysis server 13 (step E2 in Figure 2).

More specifically, the server 13 may then, firstly, identify the tasks assigned to the user during the conversation, and secondly, utilize and perform these tasks. Performs a semantic analysis of the representation sent to server 13 to isolate relevant terms that can be used to identify high services (step E31 in Figure 2).

For example, if server 13 receives all the expressions of the conversation between the user, Matthew and Alan, then server 13 says "I need to call him", "I buy them online." , "I will send you SMS (Short Message System)", and "I will send you a route", but Matthew (User 12) has four tasks after the conversation. Is used to identify what must be done. Specifically, this user must: That is, Making a phone call, Making a purchase online, Sending SMS, It is to send a route.

This semantic analysis is also used to select services that the user can use to perform the work assigned to the user. To do this, the natural language analysis server 13 can be connected to the database 15, which we call the service semantic reference system. Reference system 15 can include a list of services that can be launched from communication system 1 of user 12 and / or executed. Each service included in the reference system can be associated with: That is, A tree structure representation, where each leaf corresponds to a word, Items that are supposed to be displayed on the instant messaging client interface when the service with which the item is associated is selected (buttons, hyperlinks, tabs that can be displayed on the instant messaging client interface). , Or any other graphic element), Metadata with a set of parameters required to launch a service from communication system 1.

Each leaf of the tree structure representation can correspond to a semantic term that can have an associated base score. This base score represents the degree of relevance of tree-structured terms to the service to which the tree-structure is associated. Such expressions allow for ontological analysis.

Once the work that the user has to perform is identified, the natural language analysis server 13 then includes, for each of the tree-structured terms, in the representation in which the analysis was used to identify the work. Adapted to perform comparisons between the terms used (step E32 in Figure 2).

More specifically, for each expression used to identify a job, a service is selected from a comparison of the terms used in that expression with a plurality of tree-structured terms. At the beginning of this comparison, each service has a score of 0. Then, when a term in a structure matches a term in the expression, the score of the service for the tree structure with the term matching the separated term is the base score associated with the term in the structure. Is only incremented.

Thanks to the use of an ontology that allows a link between a separated term and at least one of the structural terms, there is a correspondence with that term of the separated term and structure. However, they do not have to be completely similar. In fact, for example, it is possible for the separated term "news" (canonical form "news") to match the structural term "information" through the advantageous use of ontology to describe the relationship between these terms. Is.

In this way, for each representation that identifies the work to be performed, server 13 selects the service with the highest overall score. The service selected in this way is potentially the best to allow user 12 to perform the job, as the service is selected based on the expressions used in the conversation. Are suitable.

In our example, four services have been selected that allow Matthew to perform the work that Matthew is in charge of. As shown in Figure 3, these different services are: That is, "Telephone service", "SMS sending service", "Online ticket reservation service", It is an "online route search service". Telephone services and SMS sending services can be performed from applications available on communication system 1. The ticket reservation service and the route search service may be executed from an internet site accessible from communication system 1.

Once jobs have been identified and related services selected, users 12 perform those jobs, knowing that the results associated with some services may determine the execution of other services. It is also of interest to identify the proper order in which those services must be launched.

To do this, the natural language analysis server 13 is adapted to perform contextual analysis of the expressions used during the conversation (step E33 in Figure 2). In our example, this analysis provides an online ticket purchasing service only if Philip does not have a ticket through the expressions "confirm" and "if he does not have them". Identify what needs to be raised. Similarly, the expression "later" indicates that an SMS must be sent to Alain when it is confirmed if Philip has a ticket or when Matthew purchases a ticket from an online purchase service.

Once the order of services has been determined, server 13 sends items and metadata related to the selected services to user 12's instant messaging client 14 (step E4 in Figure 2). The items are then displayed on the messaging client interface in the specified order (step E5 in Figure 2). This display may take the form of status chart 101.

Advantageously, the user 12 launches some services from the communication system 1 from the instant messaging client 14 of the user 12, which changes the order of the proposed services or is selected, and / Or it can be replaced with another service that is likely to be executed.

Advantageously, User 12 can also trigger the launch and execution of these different services. In that case, these services will be launched and executed in sequence, in the order established in the sequence, using the metadata that would have been sent.

The results obtained by completing each service can be reused to launch services that are scheduled to complete in a later order.

In the embodiments described above, the choice of service was made based solely on the expressions used during the conversation. However, in the list of services present in reference system 15, some services are different available using similar or equal internet sites or on communication system 1 of user 12. It can be run using an application.

For example, a concert ticket reservation service can be run from a theater internet site, from a music group site, or through a music direct sales chain site. Similarly, interlocutors can also be contacted by telephone using traditional telephone services or via voice-over IP type services.

Therefore, it may be of interest to improve the selection based on the user 12's pre-stored preferences.

To do this, the natural language analysis server 13 can also be connected to a second database 16 that contains a user profile. The user profile advantageously includes: That is, A list of favorite internet sites or pages that users 12 prefer to look up information or take action on. For each contact, including, among other things, the surname, first name, and various applications that can contact the contact (email, instant messaging, traditional phone, IP phone, conference call, etc.), as well as mentioned above. A list of user contacts, including various information required to communicate with these contacts through the application.

In the conversation example shown in FIG. 3, the first telephone call service selected from the semantic analysis of the expressions used during the conversation can also be selected based on the preference of the user 12. For example, if the user specifies in these preferences that he or she only wants to have a telephone conversation using the voice over IP application, the analysis server 13 runs from the voice over IP application. Select a possible telephone service.

Moreover, if telephone contact details regarding Philip are included in the user preferences, the metadata associated with the selected telephone service can be advantageously complemented by these details. In this way, the user does not have to enter further information to launch the telephone service from the instant messaging client 14.

Similarly, it is possible for the analysis server to examine the pre-stored preferences of user 12 and select a ticket reservation service that is specifically appreciated by user 12. For example, the server 13 can select the URL of the site of the music direct sales store stored in the favorites of the user 12.

Therefore, by using the database 16 that contains the user preferences, it is possible to customize the services provided to the user 12.

Various features and advantages of the present invention have been described in the context of conversations made from instant messaging clients, but the invention is not limited to this type of conversation. The present invention can also be applied, among other things, when the conversation is a telephone conversation.

When the conversation is a voice conversation made from the telephone terminal device 10, the communication system 1 records the conversation and sends this record to the natural language analysis server 13 at the end of the conversation. The natural language analysis server 13 then performs semantic analysis, service selection, context analysis, and service stage order identification, as described above.

Furthermore, the establishment of a telephone conversation can also trigger a display application (not shown) in communication system 1. The analysis server 13 is then adapted to send the identified service sequence to the application responsible for displaying this sequence on the graphic interface of communication system 1. This display is made, for example, on the graphic interface of computer 11 of communication system 1.