TECHNICAL FIELD

The invention relates to predicting a new network affiliation of a churned user of a telecommunication network. More particularly, the invention relates to a method, and computer program product for investigating the behaviour of a user in a telecommunication network as well as to a behaviour investigating arrangement and a telecommunication network comprising a behaviour investigating arrangement.

BACKGROUND A user of a telecommunication network, such as a mobile communication network, is in many instances a subscriber who subscribes to the use of the telecommunication network. If the user is not satisfied with the services, then he or she may end the subscription. The user does in this case become a churner.

There exist numerous causes/reasons for a subscriber to churn out of a telecom operator's subscription. It is not always possible for the operator to infer the reason behind it. It is only possible to speculate about it unless the subscriber has churned graciously giving a reason why he or she exits. However, if the churning is ungraceful, i.e. when no reason is given for the exit, it may still be of interest for the operator to find out why the user leaves and perhaps also if he or she leaves in favour of another operator.

It may thus be of interest to find out if the user changes network affiliation.

One way of predicting a new network affiliation is described in US

2012/0231781. This document describes the identifying of a new network affiliation of a churned subscriber that has moved from an old network operator to a new network operator. The identification is done using call usage attributes of the churned user. For example, if the churned user calls the same set of subscribers with his new contact number, then it is possible to predict that a new contact number calling this set, may be the churned customer's new number. The prediction may in this case be made by analysing the call graph properties.

This works well if the phone numbers contacted by the user belong to users of the old telecommunication network. However, this is not always the case.

It would therefore be of interest to use an alternative way of predicting a new network affiliation of the churned user.

SUMMARY

One object of the invention is thus to predict a new network affiliation of a churned user.

This object is according to a first aspect achieved by a behaviour

investigating arrangement in a telecommunication network. The behaviour investigating arrangement comprises a processor acting on computer instructions whereby the behaviour investigating arrangement is operative to

obtain telecommunication network data of a user being affiliated with the telecommunication network,

analyse the telecommunication network data with regard to at least one social community site visited by the user, and

investigate the same social community sites for identifying candidates for the user, determine an identity of the user within at least one social community site based on the analysis and the investigation.

In case the user ends the affiliation with the telecommunication network the behaviour investigating arrangement is further operative to

analyse the activities of the user on the at least one social community site based on the user identity, and

predict a new network affiliation of the user based on the analysed activities at the social community. This object is according to a second aspect also achieved by a

telecommunication network comprising a behaviour investigating arrangement. The behaviour investigating arrangement comprises a processor acting on computer instructions whereby the behaviour investigating arrangement is operative to

obtain telecommunication network data of a user being affiliated with the telecommunication network,

analyse the telecommunication network data with regard to at least one social community site visited by the user,

investigate the same social community sites for identifying candidates for the user, and

determine an identity of the user within at least one social community site based on the analysis and the investigation.

In case the user ends the affiliation with the telecommunication network the behaviour investigating arrangement is further operative to

analyse the activities of the user on the at least one social community site based on the user identity, and

predict a new network affiliation of the user based on the analysed activities at the social community site. The object is according to a third aspect achieved through a method for investigating the behaviour of a user in a telecommunication network. The method is performed in a behaviour investigating arrangement in the telecommunication network and comprises:

obtaining telecommunication network data of a user being affiliated with the telecommunication network,

analysing the telecommunication network data with regard to at least one social community site visited by the user,

investigating the same social community sites for identifying candidates for the user, and

determining an identity of the user within at least one social community site based on the analysis and the investigation.

In case the user ends the affiliation with the telecommunication network, the method further comprises

analysing the activities of the user on the at least one social community site based on the user identity, and

predicting a new network affiliation of the user based on the analysed activities at the social community site.

The object is according to a fourth aspect achieved through a computer program product for investigating the behaviour of a user in a

telecommunication network. The computer program product is provided on a data carrier and comprises computer program code which when run in a behaviour investigating arrangement in the telecommunication network, causes the behaviour investigating arrangement to:

obtain telecommunication network data of a user being affiliated with the telecommunication network,

analyse the telecommunication network data with regard to at least one social community site visited by the user,

investigate the same social community sites for identifying candidates for the user, and

determine an identity of the user within at least one social community site based on the analysis and the investigation. In case the user ends the affiliation with the telecommunication network the computer program code further causes the behaviour investigating arrangement to

analyse the activities of the user on the at least one social community site based on the user identity, and

predict a new network affiliation of the user based on the analysed activities at the social community site.

The invention according to the above-mentioned aspects has a number of advantages. It allows a new network affiliation of a churned user to be predicted using the social community information of a user, which enhances the reliability of the prediction. The information may also be used for a number of associated tasks. It maybe used n order to find out the root causes of a churn and also allows the operator to analyse competitors. The knowledge of the new network affiliation may also be used by the operator in recovering any outstanding amount from the churned out user.

In an advantageous variation of the first aspect, the behaviour

investigating arrangement, when investigating the same social community sites, is further configured to investigate user activities at the at least one social community site.

In a corresponding variation of the third aspect, the investigating of the same social community sites comprises investigating user activities at the at least one social community site.

The telecommunication network data may comprise session data records relating to communication sessions between the telecommunication network and the at least one social community site. According to a further variation of the first aspect, the behaviour investigating arrangement when investigating the at least one social community site is operative to investigate changes in status of at least one candidate user of the social community site.

According to a corresponding variation of the third aspect, the

investigating of the at least one social community site comprises investigating changes in status of at least one candidate user of the social community site.

The telecommunication network data may also comprise session data records relating to communication sessions between the user and friends of the user. According to yet another variation of the first aspect, the behaviour investigating arrangement is further operative to identify friends of the user via the session data records and when investigating the at least one social community site is further operative to investigate friends of the candidate user.

According to a corresponding variation of the third aspect, the method further comprises identifying friends of the user via the session data records and the investigating of at least one social community site further comprises investigating friends of the candidate user.

According to a further variation of the first aspect, the behaviour investigating arrangement is further operative to rank the social community sites based on user activity at the sites. According to a corresponding variation of the third aspect, the method further comprises ranking the social community sites based on user activity at the sites. According to a further variation of the first aspect, the behaviour investigating arrangement is further operative to match

telecommunication network data to data of a social community site when determining an identity of the user at a social community site.

According to a corresponding variation of the third aspect, the determining of an identity of the user at a social community site comprises matching telecommunication network data to data of a social community site.

According to yet a further variation of the first aspect, the behaviour investigating arrangement is further operative to, when matching telecommunication network data to data of a social community site, compare user activities and friends in the telecommunication network with activities and friends of the candidate users and select the candidate user as the user, for which user activities and friends best match the user activities and friends of the user in the telecommunication network.

According to a corresponding variation of the third aspect, the matching comprises comparing user activities and friends of the user in the telecommunication network with activities and friends of the candidate users and selecting the candidate user as the user for which user activities and friends best match the user activities and friends of the user in the telecommunication network.

According to another variation of the first aspect, the behaviour

investigating arrangement is further operative to apply data about activities and friends in a similarity function when comparing user activities and friends of the user with the activities and friends of the candidate users. According to a corresponding variation of the third aspect, the comparing of user activities and friends of the user with the activities and friends of the candidate users comprises applying data about activities and friends in a similarity function.

According to yet another variation of the first aspect, the behaviour investigating arrangement is further operative to, when determining an identity of the user and in the case of a tie between candidate users, compare the candidate user with a user of at least one social community site for which a match has been made and select the candidate user at the investigated social community site that best matches the determined user of the other social community site.

According to a corresponding variation of the third aspect, the method further comprises, in the case of a tie between candidate users when determining an identity of the user, comparing the candidate users with a user of at least one social community site for which a match has been made and select the candidate user at the investigated social community site that best matches the determined user of the other social community site.

According to yet a further variation of the first aspect, the behaviour investigating arrangement when analysing the activities of the user on the at least one social community site is operative to analyse user activities in relation to telecommunication as well as analyse the similarity between a behaviour of the user according to these user activities and the behaviour of the user as manifested in session data records in the telecommunication network.

According to a corresponding variation of the third aspect, the analysing of the activities of the user on the at least one social community site comprises analysing user activities in relation to telecommunication as well as analysing the similarity between a behaviour of the user according to these user activities and the behaviour of the user as manifested in session data records in the telecommunication network.

According to still a further variation of the first aspect, the behaviour investigating arrangement is further operative to obtain a communication identifier of a further telecommunication network to which the user is predicted as being affiliated and provide the communication identifier for use in a traffic handling function in the telecommunication network for performing a traffic handling activity in relation to traffic intended for the user

According to a corresponding variation of the third aspect, the method further comprises obtaining a communication identifier of a further telecommunication network to which the user is predicted as being affiliated and providing the communication identifier for use in a traffic handling function in the telecommunication network for performing a traffic handling activity in relation to traffic intended for the user,

It should be emphasized that the term "comprises/comprising" when used in this specification is taken to specify the presence of stated features, integers, steps or components, but does not preclude the presence or addition of one or more other features, integers, steps, components or groups thereof. BRIEF DESCRIPTION OF THE DRAWINGS

The invention will now be described in more detail in relation to the enclosed drawings, in which:

fig. l schematically shows a user communicating with various social community sites via a mobile communication network,

fig.2 shows a block schematic of a first way of realizing a user behaviour investigating arrangement in the mobile communication network, fig. 3 shows a block schematic of a second way of realizing the user behaviour investigating arrangement in the mobile communication network,

fig. 4 shows a flow chart of method steps in a method for investigating the behaviour of the user in the mobile communication network according to a first embodiment,

fig. 5 shows a flow chart of method steps in a first part of a method for investigating the behaviour of the user in the mobile communication network according to a second embodiment,

fig. 6 shows a flow chart of method steps in a second part of the method for investigating the behaviour of the user in the mobile communication network according to the second embodiment, and

fig. 7 shows a computer program product comprising a data carrier with computer program code for implementing the functionality of the t user behaviour investigating arrangement.

DETAILED DESCRIPTION

In the following description, for purposes of explanation and not limitation, specific details are set forth such as particular architectures, interfaces, techniques, etc. in order to provide a thorough understanding of the invention. However, it will be apparent to those skilled in the art that the invention maybe practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known devices, circuits and methods are omitted so as not to obscure the description of the invention with unnecessary detail.

Fig. l schematically shows a telecommunication network, which is exemplified by a mobile communication network MN 12 comprising a base station BS 12 connected to a serving gateway SGW 16. The serving gateway

16 is in turn connected to a PDN Gateway PGW 18, where PDN is an acronym for Packet Data Network. In the mobile communication network 12 there is also a customer relations manager (CRM) device 20 as well as a user behaviour investigating arrangement 22. The arrangement 22 is connected to the SGW 16, the CRM 20 and the PGW 18. It should be realized that the mobile communication network 12 may comprise several more devices. There may also be more devices of the same type, such as PGWs, SGWs and base stations. The mobile communication network may furthermore be a network allowing Internet connectivity such as Long Term Evolution (LTE) or Wideband Code Multiple Access (WCDMA). Aspects of the invention will in the following be described in relation to the mobile communication network 12. However, the telecommunication network is not limited to mobile communication networks, but may for instance be a Public Switched Telecommunication Network (PSTN).

The base station 14, which is often termed eNodeB or just NodeB, is furthermore provided in a part of the mobile communication network 12 termed access network, while the other devices are provided in a part of the mobile communication network 12 termed a core network.

A user Ui of the mobile communication network 12 is furthermore equipped with a terminal 10, often termed a mobile station MS, via which he or she may communicate with other users via the mobile

communication network 12. The user Ui may more particularly share information with other people via so called, social communicate sites, such as Facebook, Linkedln or Twitter. As an example three such sites are shown as being connected to the PGW 18 for such social community information sharing. The link between the PGW 18 and such a site is typically provided via the Internet. There is here a first social community site SCSi 24, a second social community site SCS2 26 and a third social community site SCS3 28. The numbers are mere examples. There may of course be more as well as fewer sites. Fig. 2 shows a block schematic of a first way of realizing the user behaviour investigating arrangement UBIA 22. It maybe provided in the form of a processor PR 30 connected to a program memory M 32. The program memory 32 may comprise a number of computer instructions

implementing the functionality of the user behaviour investigating arrangement 22 and the processor 30 implements this functionality when acting on these instructions. It can thus be seen that the combination of processor 30 and memory 32 provides the user behaviour investigating arrangement 22.

Fig. 3 shows a block schematic of a second way of realizing the user behaviour investigating arrangement 22. The user behaviour investigating arrangement 22 may comprise a Network Data Analyser NDA 33, a Social Community Site Investigator SCSI 34, a User Identity Mapper UIM 35 and a Network Affiliation Predictor NAP 36.

The elements in fig. 3 may be provided as software blocks for instance as software block in a program memory, but also as a part of dedicated special purpose circuits, such as Application Specific Integrated

Circuits(ASICs) and Field-Programmable Gate Arrays (FPGAs). It is also possible to combine more than one element in such a circuit.

As mentioned earlier the user Ui of the mobile communication network 12, which user Ui may be a subscriber, may not be satisfied with the services provided by the operator of the mobile communication network

12, and may in this case end the subscription. The user Ui does in this case become a churner.

There exist numerous causes/reasons for a subscriber to churn out of a telecom operator's subscription. It is not always possible to infer the reason behind it. One can only speculate about it unless the subscriber has churned graciously giving a reason why he or she exits. However, if the user churns ungracefully giving no reason for the exit, it may be of interest for the operator to find out why the user leaves and perhaps also if he or she leaves in favour of the operator of another mobile communication network.

In most cases of subscriber churning out, there will not be a graceful churn-out. In this case it may be possible to find the subscriber's new operator or his/her new communication identifier with the help of social media. A communication identifier may in this case be a Mobile Station International Subscriber Directory Number (MSISDN), which is typically the phone number of the subscription.

Aspects of the invention are directed towards predicting a new network affiliation, which predicting is based on a correlation between the telecommunication network that the user has left and social media information in order to identify the user at various social community sites and then use this knowledge in order to find out the new network affiliation. Aspects of the invention are thus directed towards tracking the social profile of the user Ui, which could be investigating blogs, forums, social networking sites and so on. Then, if the user Ui churns, the new network affiliation may be predicted based on social media information and possibly also based on usage information.

Now a first embodiment will be described with reference also being made to fig. 4, which shows a flow chart of method steps in a method for investigating the behaviour of the user Ui in the mobile communication network 12, which method is being performed by the user behaviour investigating arrangement 22. In order to be able to determine a new network affiliation after a churn it may be necessary to look at data in the mobile communication network before the user has churned or left the network 12. Therefore the network data analyser 33 of the user behaviour investigating arrangement 22 obtains mobile communication network data of the user Ui when the user Ui is affiliated with the mobile communication network 12, step 38. The mobile communication network data, in the following termed network data, may comprise session specific data such as data about actual communication sessions between the user and the social community sites using the mobile station 10 and the gateways 16 and 18. Such session specific data may be collected in session data records (SDR) and then be used to find out when and to which social community sites the mobile terminal of the user connects to. At times it may also be possible to find out user identity data from such session data records, such as if a user identity at a site is attached to a Uniform Resource Locator (URL) used for connecting to the site. SDRs may also be analysed in order to find out which friends the user Ui contacts as well as how frequent and how long such sessions with friends are. The phone numbers of the friends may then be located through the session data records, which session data records relate to communication sessions between the user and friends of the user. The network data analyser 33 may thus identify friends of the user via the session data records. The customer relations management (CRM) data may also comprise customer relations data, such as the types of services subscribed to by the user, the pricing as well as the name of the user Ui him- or herself. CRM data may also comprise basic demographic data of the user and complaints data. CRM data may then be kept at the CRM device 20. The name may however also be provided in other places in the mobile communication network 12, such as in a Home Location Register

(HLR). The network data analyser 33 may communicate with the SGW 16 and/or the PGW 18 in order to obtain SDRs and with the CRM device 20 in order to obtain CRM data. The network data analyser 33 furthermore analyses the network data with regard to social community sites being visited by the user, step 40. It thus analyses the network data with regard to at least one social community site 24, 26 and 28 visited by the user Ui. The network data analyser 33 may thus find out which sites have been visited at which times and if any useful data has been used when the user has logged into theses sites. Such information may be obtained through investigating the session data records and/ or the CRM data. Other possible information that may be obtained, for instance from the CRM device 20 or from the HLR, is a name of the user Ui.

At the same time the social community site investigator 34 investigates the same social community sites for identifying candidates for the user Ui at these sites, step 42. The investigation may involve investigating changes at the sites, such as which users are being logged in at the time of the above mentioned session specific network data as well as what activities they are involved in. It thus investigates user activities at the social community sites and perhaps changes in status of at least one candidate user of the social community sites. The social community site investigator 34 may obtain name information of such users being logged in at these sites. This name information may be compared with name information of the user Ui. The investigation may also comprise an investigation of the friends of the candidate user.

Thereafter the user identity mapper 35 determines an identity of the user Ui within at least one of the investigated social community sites based on the analysis and the investigation, step 44. If there is only one candidate at a site, this may be directly selected. If there are more then one candidate for a site, the candidate that best matches the user is selected. There may thus be a matching of the data of the candidates to the network data of the user Ui. The determining of an identity of the user at a social community site may therefore be achieved through matching of mobile

communication network data to data of a social community site. The candidate determined to be the user Ui may be the candidate that has the most similar name and the most similar friends, i.e. there is a similarity l6 between the persons listed as friends on the social community sites and the friends of the user Ui according to an analysis of the session data records. The matching may thus comprise comparing user activities and friends in the mobile communication network with activities and friends of the candidate users and selecting the candidate user as the user, for which user activities and friends best match the user activities and friends of the user in the mobile communication network. In the case of Facebook the user identity may be a part of a SDR and in this case the candidate having this user identity at the site may be directly selected.

This mapping may be an ongoing process as long as the user is affiliated to the mobile communication network 12. The user Ui may leave social communities and join others and the identities of the user at these sites determined. The network affiliation predictor 36 then investigates if the user Ui has ended the affiliation with the mobile communication network 12, and if he or she has not, step 46, then the investigation is ended or continued with regard to new sites visited, step 48.

However, if the user Ui has ended the affiliation, step 46, for instance through ending the subscription, the network affiliation predictor 36 then analyses the activities of the user on the social community sites 24, 26 and 28, step 50, and predicts a new network affiliation of the user based on these activities, step 52. In case the user ends the affiliation with the mobile communication network 12, the network affiliation predictor 36 thus analyses the activities on the sites based on the respective user identities and thereafter predicts a new network affiliation. It may for instance analyse social community site data such as postings, tweets and likes made by the user Ui at the various sites and determine the new network affiliation based on these postings, tweets and likes. It is also possible that the similarity between this social community site data and network data is employed as a basis for the prediction, where the network data may here comprise calling behaviour data and CRM data, where the calling behaviour data is obtained from an analysis of the previously mentioned SDRs. The user may furthermore have provided a new connectivity or communication identifier such as a new MSISDN openly at the site, from which the new network affiliation may be obtained. The user may also or instead have provided a posting where the new network affiliation is mentioned. In this case the new network affiliation may also be directly obtained.

It can thereby be seen that information provided by the user Ui at various social community sites, such as Facebook, Linkedin and Twitter, is used for determining a new network affiliation.

It may thus be possible to use the social media information of a user to find his/her new network affiliation. For instance, the user might update his/her social media profile (Twitter, Facebook, Linkedin, etc.) with a new contact number or communication identifier. Motivation to do this lies in the fact that social networking sites such as Facebook, Linkedin, Twitter, and Xing have been increasingly gaining in popularity. The average user may check his favourite community site several times a day and a large part of the mobile phone users use their phones to access the social community sites.

Multiple sources of information may thus be used for finding the new network affiliation/MSISDN of a churner. The information used is furthermore social media/ open information. The information may also be used in order to find out the root causes of a churn and also allows the operator to analyse competitors.

The knowledge of the new network affiliation may also be used by the operator in recovering any outstanding amount from the churned out subscriber. l8

Now a second embodiment will be described with reference first being made to fig. 5, which shows a flow chart of method steps in a first part of a method for investigating the behaviour of the user in the mobile

communication network.

Just as in the first embodiment, the network data analyser 33 of the user behaviour investigating arrangement 22 obtains network data of the user Ui, when the user is affiliated with the mobile communication network 12. It does more particularly obtain session data records, for instance records of sessions in which data is exchanged between the mobile communication network 12 and the various social community sites 24, 26 and 28, step 54. The information in a session data record may thus comprise information about data exchanged between the site and the mobile station 10, when the user Ui accesses a social community site. It may thus involve the user logging onto an own account at the site, performing some updates at an area of the site dedicated to this account and thereafter perhaps logging out. The session data records also comprise information of communication sessions between the user Ui and friends of the user Ui, which

communication sessions may involve telephone calls. The session data records thus only comprise session data records involving the user Ui.

The network data analyser 33 also obtains CRM data of the user Ui from the CRM device 20, step 56. Such data may comprise the name of the user Ui, different types of services that the user subscribes to as well as various rates and fees of such services.

There may be a number of social community sites which the user Ui accesses. In the following these may be considered as graphs, where the different site users are nodes of these graphs. Also the mobile

communication network 12 may be considered to be a graph and the different mobile communication network users may be considered to be nodes of this graph. In order to find out which identity the user has on the different sites and consequently also to find out which areas of these sites that are assigned to the user, the identification of user identities at the sites may thus be treated as a node mapping problem. Also in this second embodiment, the network data analyser 33 analyses the network data with regard to social community sites being visited by the user, step 57. The analysis may involve analysing demographic data in order to be able to perform a basic search for the user at various social communities. The analysis may also involve analysis of the session data records with regard to at least one social community site visited by the user Ui. By analysing the session data records, the network data analyser 33 may thus find out which sites have been visited at which times and if any useful data has been used when the user has logged into theses sites, such as a user identity. The session data records may also be used to identify friends of the user, which friends are identified via their phone numbers, such as via their MSISDN numbers.

The network data analyser 33 furthermore analyses the session data records in order to establish or determine a communication behaviour of the user Ui, step 58. The behaviour may be specified in the form of user communities, frequent callers, frequent called numbers, calling groups and higher order groups. The analysis may also involve an analysis of complaints that the user has made. Thereafter the Social Community Site investigator ranks the social community sites for the user Ui, step 59, which ranking is performed based on user activity at the sites.

Each user of the mobile communication network 12, and thus also the user Ui, may have a ranked/ weighted list of online social networks, i.e. social communities, based on simple ranking of the usage. If the user does not have a data connection with the operator, then the default popularity ranking of the site can be taken as the user's ranked/weighted list of social networks. It maybe defined formally as follows:

Weight (social community) <— Ψ (data usage information)

One possible instantiation of it could be Ψ (Freq, duration, recent) where each social community is ranked based on frequency, duration and recent usage. And the ranks can be combined using the function Ψ, which can be a linear combination of the parameters.

The social communities may thus be ranked based on the CRMs with session specifying data concerning sessions between the mobile station 10 of the user Ui and the different social community sites 24, 26 and 28. Thereafter the social community site investigator 34 investigates the social community sites. This investigation may be performed in the ranked order, where the top ranked site is examined first. The social community site investigator 34 more particularly obtains candidates at each social community site based on changes in status in different areas of these sites, and timing of activities at these sites according to the session data records, step 60. This means that it may investigate various users at a site to see which users are logged into their accounts at the site at the time at which a session data record indicates that the user Ui is connected to the same site. It is also possible to use some other data, such as if the CRM device 20 or the HLR has knowledge of the name of the user Ui, for identifying candidates. In this case it is possible to look at if any of these logged on users have names that are the same or similar to the name of the user Ui as known by the CRM device 20. The social community site investigator 34 may thus see what accounts have been logged into and out of at the times of the session data records as well as if any of the account holders have similar names as the user. In this way it is possible to obtain a number of candidates at each social community site that may be the user Ui. Furthermore only open data, i.e. data that these candidates themselves allow to be seen, is investigated at the sites.

Thereafter the user identity mapper 35 determines an identity of the user within at least one of the investigated social community sites based on the analysis and the investigation. If there is only one candidate at a site, this candidate may be directly selected. If there are more candidates, then the candidate, which best matches the user is selected. In order to select one candidate when there are several different candidates, the user identity mapper 35 may determine the similarity of friends and usage for each candidate at each social community site, step 62.

This may, as was stated earlier, be treated as a node matching problem between two (or more) social graphs. One graph is a graph comprising users of the mobile communication network 12 and every other graph represents a respective social community site. Each node in a graph representing a social community site is a user which has both attributes X (for example profile, posts/tweets/updates, etc.) and social network L (friends/followers/peer group). Similarly, the graph representing the mobile communication network captures attribute (CRM) and social network (call graph) information for each user of the mobile

communication network 12. Here the task is to map a node (representing the user Ui) in the mobile network graph to a node in a graph representing a social community. It is not necessary that all mobile users would have a profile on a social community; the idea here is to map the users who have one (as many as possible).

To find a social community identity of a user using his browsing data would not be straight forward because the session data records cannot be used to capture the identity in all the cases. For instance a Facebook identity can be found, but not twitter and Linkedln. Also a user might not have taken a data plan with that particular operator. Secondly, not all users in the social community may keep their contact information public. So it is not just a CRM vs. social community profile matching. In such cases, the social community identity may be identified using both CRM and session data.

If a user's identity can be mapped using his CRM data only, then this can be used to map his/her common friends in the mobile communication network and at the social community. Otherwise, if usage behaviour and friends' network in one or more social communities are known, this can also be used to identify a user. This again can be used to map his/her friends. The concept may be to map users confidently and then propagate the identity information for identifying his/her friends. This is continued until any more mapping is no longer possible. It is thus possible to use a heuristic based Map Method (MN, SCi, ... SCN) Input: MN - mobile communication network, SCi to SCN— social communities.

It is possible to perform a preliminary match: Here the name and demography information of the user in the mobile communication network and at the social community sites are matched. There might exist more than one match in a social community, which is handled in the subsequent steps. Secondly, there are cases like Facebook, where the user identity can be captured in the packet information. For example, a user named "sample", would have www.facebook.com/ sample in a session data record. It is not necessary that all social community identities must be available in the session data records. For example Linkedln generates a key for each session. After the preliminary match it is then possible to map a user in the mobile communication network MN to a social community site SNi which has a similarity value > Δ (Δ can be fixed empirically) Similarity = α (Θ, t) a may be a function over attributes and link (social) information, a may for example be a weighted sum of the attribute based and link based similarities. Θ may be a combined similarity value of usage information. An example usage information based similarity would be if the time of tweets (and/or) posts matches with the time of connection to the sites of Twitter and/ or Facebook according to the session records, then the similarity value is high. It is computed using a sum of hits in usage.€ is a function over similarity between each of mobile communication network

neighbours (call graphs) and social community neighbours. It can be a weighted combination, where weights are derived based on total duration of calls between two users. It is thus a mapping of persons being called by the user Ui in the mobile communication network 12 and persons indicated as friends by the candidate at the site. α(θ, O = S(0) X S(t), where S is sigmoid function which can map it between the range [0,1]. As can be seen above, the comparing of user activities and friends of the user with the activities and friends of the candidate users may comprise applying data about activities and friends of both the mobile

communication network and the social community sites in a similarity function. The activities in the mobile communication network were above exemplified by data in session records. It is also possible that CRM data comprise data of activities, such as if the user has made complaints. The User identity mapper 35 then investigates if there is a tie between top candidates, i.e. there are two or more top candidates that are equally similar to the user Ui in the mobile communication network. If there is not, step 64, then the most similar is selected, step 66.

There may thus be a matching of the data of the candidates to the network data of the user Ui. The candidate determined to be the user Ui may be the candidate that has the most similar name, the most similar use and the most similar friends, i.e. there is a similarity between the persons listed as friends on the social community sites and the contacts of the user Ui of the CRM device 20. In the case of Facebook the user identity may be a part of the network data and in this case the candidate having this user identify at the site maybe selected without any further analysis. However, further analysis may still be performed in order to confirm the identity.

However if there is a tie, step 64, then the user identity mapper 35 continues and determines similarities between the candidates and selected candidates at other social community sites, step 68. It thus investigates the similarity between the two or more candidates of a site that are tied as being most similar to the user Ui in the mobile communication network and compares these candidates with users at other social community sites having already been determined to correspond to the user Ui in the mobile communication network 12. Thus, in cases of multiple matches with a social community SCi, the similarity of the mapped entity in one or more and sometimes all other social communities may be used to this social community to break the tie.

Here the following function may be used: ¾ Wj * Sim (SCi * SCj) , where j is 1 to N except i. Wj is the weight assigned to social community SQ based on the usage information. Thereafter the candidate that is shown to be more similar to these other selected social community users, is selected to be the user Ui, step 70.

It can thus be seen that, in the case of a tie between candidate user identities the user identity mapper 35, when determining an identity of the user, compares the candidate users with a user of at least one social community site for which a match has been made and selects the candidate user at the investigated social community site that best matches the determined user of the other social community site. In this way it is possible to iterate through the steps of determining similarities between candidates at a site and handling of ties until there is a convergence (all the nodes are mapped or a fixed number of iterations have been performed). This mapping may be an ongoing or recurring process for a node as long as the user representing the node is affiliated to the mobile communication network 12. The user may leave social communities and join others. The user identity mapper 35 may thus keep on determining the user identity at any new social community that the user Ui joins as long as he or she is affiliated with the mobile communication network 12.

The operation of the user behaviour investigating arrangement 22 according to the second embodiment when the user has left the mobile communication system 12 will now be described with reference being made to fig. 6, which shows a flow chart of method steps in a second part of the method for investigating the behaviour of the user in the mobile communication network.

Fig. 6 shows a flow chart of method steps that are used when the user is no longer affiliated to the mobile communication network. In this case there are no longer any new sessions in the mobile communication network 12 between the user and the social community sites and consequently no session data records of such sessions that may be used, but as the user has been identified at various social communities, the user activities at these may be analysed in order to predict a new network affiliation. In this regard it may also be possible to consider the communication behaviour of the user as well as CRM data of the mobile communication network.

The network affiliation predictor 36 thus analyses the user activities of the social community sites 24, 26 and 28 and predicts a new network affiliation based on these activities. More particularly it analyses the user activities at these sites with regard to telecommunication, step 72. The analysis of user activities at the sites may involve analysis of postings, tweets and likes made by the user Ui at the various sites. The network affiliation predictor 36 may also look at the activities of the friends of the user at these sites, step 74. It may also look at the user communication behaviour, step 76, as well as the CRM data, step 78. The network affiliation predictor 36 then predicts the new network affiliation, step 80. In case the new network affiliation can be directly found out through the user activities, for instance through the user specifying a further network, the use of which the user now subscribes to or a new MSISDN, then this information may be directly used for determining the new network affiliation. However, if there is no such direct indication, the network affiliation predictor 36 may predict the new network affiliation based on the similarity of the behaviour at the site and the user communication behaviour. It is thus possible that the similarity between the behaviour at a social community site, the mobile communication network behaviour and CRM data is employed as a basis for the prediction. As can thereby be seen the network affiliation predictor 36 may analyse the similarity between a behaviour of the user according to the user activities at the social community sites and the behaviour of the user as manifested in the session data records of the mobile communication network 12.

The network affiliation may be captured using a set of monitored parameters of that particular user, a sample of which are given in table 1.

Table 1

These parameters, which are thus indicative of the social community behaviour of the user Ui, can be used in addition to network parameters that have previously been monitored. Sample of network parameters is given below, which thus provide information about the user behaviour in the mobile communication network. Mobile network parameters may be computed for each node as mentioned in Table 2 below. One concept that could be used in the mapping is to assume that a churned user might exhibit similar (with regard to a threshold) calling behaviour with a new number and thus the behaviour determined in the mobile communication network may be assumed to be valid for the further mobile communication network obtained as a result of the prediction.

Table 2 With these two types of information (network behaviour and social community behaviour along with CRM), it would be possible to predict a user's new network affiliation and/ or MSISDN confidently than using one of the sources only. It may also be possible to tune the thresholds using training data which can be labelled manually or using a data set which has gracefully churned customers.

It is furthermore possible that a communication identifier of the user in the predicted further communication network is used in some way. It may for instance be returned as a service to people trying to the call the user Ui in the mobile communication network. It is also possible to forward a call to the user in the further mobile communication network, i.e. forward an incoming call to the further mobile communication network using the new communication identifier. These activities may then be performed for instance by the PGW 18 after being instructed by the arrangement 22 or some other entity in the network 12 where the arrangement has stored information of the further network and the new number, such as in a Home Location Register (HLR). The network affiliation predictor 36 may thus obtain a communication identifier of the further mobile

communication network to which the user is predicted as being affiliated and provide this communication identifier for use in a traffic handling function in the mobile communication network for performing a traffic handling activity in relation to traffic intended for the user.

The user behaviour investigating arrangement 22 may, as was mentioned initially, be provided in the form one or more processors with associated program memories comprising computer program code with computer program instructions executable by the processor for performing the functionality of the user behaviour investigating arrangement.

The computer program code of a user behaviour investigating

arrangement may also be in the form of computer program product for instance in the form of a data carrier, such as a CD ROM disc or a memory stick. In this case the data carrier carries a computer program with the computer program code, which will implement the functionality of the above-described user behaviour investigating arrangement. One such data carrier 82 with computer program code 84 is schematically shown in fig. 7.

Furthermore the network data analyser of the behaviour investigating arrangement maybe considered to form means for obtaining

telecommunication network data of a user being affiliated with the telecommunication network and means for analysing the

telecommunication network data with regard to at least one social community site visited by the user. The network data analyser may furthermore be considered to form means for identifying friends of the user via the session data records. The social community site investigator may in turn be considered to form means for investigating the same social community sites for identifying candidates for the user. The means for investigating the same social community sites may further be considered to form means for

investigating user activities at said at least one social community site. The means for investigating the same social community sites may further be considered to form means for investigating changes in status of at least one candidate user of the social community sites. The means for investigating the same social community sites may furthermore be considered to form means for investigating friends of the candidate user. The social community site investigator may furthermore be considered to form means for ranking the social community sites based on user activity at the sites.

The user identity mapper may in turn be considered to form means fpr determining an identity of the user within at least one social community site based on the analysis and the investigation. The means for

determining an identity of the user at a social community site may furthermore be considered to comprise means for matching

telecommunication network data to data of a social community site. The means for matching telecommunication network data to data of a social community site may furthermore comprise means for comparing user activities and friends in the telecommunication network with activities and friends of the candidate users and selecting the candidate user as the user, for which user activities and friends best match the user activities and friends of the user in the telecommunication network. The means for comparing user activities and friends of the user with the activities and friends of the candidate users may furthermore be considered to form means for applying data about activities and friends in a similarity function. Furthermore, the means for determining an identity of the user may in the case of a tie between candidate users, be considered to further comprise means for comparing the candidate users with a user of at least one social community site for which a match has been made and select the candidate user at the investigated social community site that best matches the determined user of the other social community site. The network affiliation predictor may finally be considered to form means for, in case the user ends the affiliation with the telecommunication network, analysing the activities of the user on the at least one social community site based on the user identity, and means for predicting a new network affiliation of the user based on the analysed activities at the social community sites. The means for analysing the activities of the user on the at least one social community site maybe considered to be means for analysing user activities in relation to telecommunication as well as means for analysing the similarity between a behaviour of the user according to these user activities and the behaviour of the user as manifested in session data records in the telecommunication network. The network affiliation predictor may also be considered to comprise means for obtaining a communication identifier of the further telecommunication network to which the user is predicted as being affiliated and provide the

communication identifier for use in a traffic handling function in the telecommunication network for performing a traffic handling activity in relation to traffic intended for the user. While the invention has been described in connection with what is presently considered to be most practical and preferred embodiments, it is to be understood that the invention is not to be limited to the disclosed embodiments, but on the contrary, is intended to cover various

modifications and equivalent arrangements. Therefore the invention is only to be limited by the following claims.