METHOD, DEVICE AND COMPUTER PROGRAM FOR MONITORING TELEPHONE CALLS

Description

Field of invention

[0001] The invention relates to a method, a device and a computer program for monitoring telephone calls in a telephony service network. background

[0002] In telephony service networks, subscribers can establish voice connections with each other. Each subscriber line in a telephony service network is assigned a subscriber identifier (TNI) known within the network. A TNI can be a telephone number, which, in addition to a subscriber number, may include an area code, a network prefix, and / or a country code. A TNI can also be an identifier in a digital telephony service network, such as a Session Initiation Protocol (SIP) address, an International Subscriber Identity (IMSI) of a mobile network, or another unique identifier for a subscriber line. Telephony services can be provided in both analog networks, often referred to as Public Switched Telephone Networks (PSTN) or Plain Old Telephone Services (POTS), and digital networks. The latter digital networks can be fixed-line networks (e.g.,Digital networks include Integrated Services Digital Network (ISDN), Session Initiation Protocol (SIP) networks, etc., as well as mobile networks (e.g., 3GPP, 3GPP2, GSM, UMTS, LTE, 5G NR) and IP Multimedia Subsystem (IMS) networks. While traditional telephony service networks are generally circuit-switched, digital networks, at least in lower network layers, often operate on a packet-switched basis, for example, using the Internet Protocol (IP).

[0003] To make a telephone call, a calling party (hereinafter referred to as the "originator") establishes a telephone connection using the TNI of a receiving party (hereinafter also referred to as the "receiver"). A telephony service network can also determine the originator's TNI or transmit it to the receiver's terminal for display. The receiver can then, for example, decide whether to accept the call based on the originator's TNI being displayed.

[0004] Regardless of the technology used by a telephony service network, accepting a telephone call requires that a certain, possibly variable, transmission bandwidth be provided and maintained by the telephony service network for the duration of the call.

[0005] The load on a telephony network can be measured, for example, in Erlangs, which represent the number of telephone calls of a specific duration per unit of time. Every telephony network has a finite load limit. Furthermore, in digital networks (including mobile networks), the bandwidth available for telephony services must be shared with other digital services (e.g., for data transmission for web, video, documents, etc.). Therefore, network operators need to monitor the telephone calls routed through the network to keep the load low and prevent network congestion.

[0006] One way to reduce the load on a telephony network is to not forward unwanted calls, or not forward them immediately. Not forwarding can involve rejecting the call, so the originator has no way to reach the recipient. Alternatively, a warning can be displayed on the recipient's device indicating that it is a potentially unwanted call, thus preparing the recipient for the call and increasing their likelihood of rejecting it. Another option is not forwarding the call to voicemail. Both the aforementioned warning and voicemail forwarding can, at least statistically, reduce the Erlang load on the telephony network.Forwarding a call to voicemail can also include verification of the originator via voice or keystroke. This can, for example, prevent calls from automated telephone systems (phone bots, robocallers).

[0007] An unwanted phone call can be a call made without the recipient's explicit prior consent or against their express wishes, a call perceived by the recipient as disruptive or inappropriate, or one related to illegal activities such as fraud, identity theft, or extortion. Examples include telemarketing calls, spam calls, or stalking calls.

[0008] Telephone calls can be filtered using appropriate technical devices, which are installed, for example, in a participant's terminal device or in the telephony service network; that is, it is determined whether a telephone call is forwarded to the recipient or not.

[0009] Call filters based on a so-called whitelist or blacklist mechanism are well-known. This means that calls originating from a subscriber identifier stored in a whitelist are always forwarded (possibly with or without further verification), while calls originating from a subscriber identifier stored in a blacklist are blocked, i.e., not forwarded. It is also known that specific subscriber identifiers, such as telephone numbers, can be assigned different ringtones on the recipient's device or even icons on the recipient's device display.

[0010] BALASUBRAMANIYAN VA ET AL: "CallRank: Combating SPIT Using Call Duration, Social Networks and Global Reputation" refers to a call screening framework that uses a "reputation score" of a caller against a called party. The reputation score is based on the duration of previous calls between the caller and the called party. The longer the calls, the better the reputation score. WO2010034516 discloses a credit score that is determined between pairs of callers and called parties. The document indicates that the credit score is determined for call verification (or text chat verification) purposes. The credit score depends on the historical call duration between the parties. The credit need not be symmetrical; that is, it can differ depending on the direction of the call.

[0011] Document US 8,577,002 B2 discloses a system and procedure for verifying telephone calls to a subscriber number, which includes a review list of telephone numbers ("greylist") assigned to suspicious callers. Telephone numbers on the review list are further assigned a verification score. The procedure also includes rules defining the circumstances under which a telephone number is added to or removed from the review list, as well as rules concerning changes to the verification score. Data relating to the calls is received and processed by the subscribers, and the telephone number is verified according to the rules and based on the assigned verification score. Subscribers can add any telephone number to a personal list of telephone numbers to be blocked ("blacklist") or a list of telephone numbers to be accepted ("whitelist").

[0012] A verification value assigned to a caller can lead to calls from some originators being erroneously not forwarded, even though at least some of these calls might be considered desirable by one or more recipients, while the caller is undesirable to other potential recipients. While it is known that a user can create an individual whitelist or blacklist of users, entering and maintaining such lists requires considerable interaction and effort on the part of the user. Therefore, while known methods for filtering phone calls can reduce the load on a telephony network, their imprecise filtering effect makes them of limited use to users.Furthermore, such methods only provide a static snapshot of a user's ranking compared to other participants and require continuous data maintenance to keep pace with a changing environment. Therefore, there is a need for methods, devices, and / or computer programs that overcome at least some of these disadvantages. Summary of the invention

[0013] Aspects of the invention include a method according to claim 1.

[0014] The process is based on a list of subscriber identifiers (VNIs), which is stored and managed, for example, in a telephony service network. This list may include some or all of the VNIs of subscribers registered as such in the telephony service network. Furthermore, a telephony service network can receive calls from other telephony service networks (e.g., networks of other network operators and / or networks from other countries) with the corresponding VNIs. Such foreign network VNIs can also be part of the VNI list. Each VNI in the list is associated with an interaction set of subscriber identifiers from outgoing and / or incoming telephone calls.For example, for a TNI A from the list, a TNI B should always be included in the interaction set assigned to A if an interaction in the form of an outgoing call from A to B or an incoming call from B to A is detected, and B was not (yet) assigned to the interaction set of TNI A. The interaction set of a TNI (e.g., TNI A) from the list thus contains those TNIs (e.g., TNI B) with which at least one interaction in the form of an incoming or outgoing call has occurred. In other words, if B is assigned to the interaction set of A, then a direct telephone call relationship exists between A and B.

[0015] The entirety of interaction set assignments to each TNI from the list ultimately defines the edges of a social network graph. This can also be represented as a tree diagram, where the individual nodes of the tree correspond to the participants' TNIs. Thus, it is possible to evaluate calls made by TNIs in the telephony service network using social network analysis. For example, not only can individual participants in the telephony service network be evaluated, but also direct and, in particular, indirect telephone call relationships can be considered. Continuing the example above, TNI B could also be assigned to the list of participant identifiers, where B, in turn, is assigned to an interaction list. For example, TNI C could be part of B's ​​interaction list.While there is a direct telephone call relationship between B and C, there is an indirect telephone call relationship between A and C. Indirect telephone call relationships can be significant when, for example, a call is made from A to C.

[0016] Furthermore, each TNI in the interaction set is assigned an interaction metric of that TNI in the list, relative to the respective other TNI in the interaction set. Continuing the example above, TNI A from the list is thus assigned an interaction metric – hereinafter referred to simply as "metric" – of A relative to B.

[0017] A metric can be assigned to a TNI (Telephone Network Interaction), a phone call relationship in the aforementioned social network graph, a group of TNIs, or a phone call. The value of a metric can characterize trustworthiness. For example, the range of values ​​for a metric is floating-point numbers (with a decimal point) between 0.0 and 1.0, where a metric of 0.0 corresponds to no or minimal trustworthiness, and 1.0 corresponds to maximum or the highest possible trustworthiness. Any other technically meaningful mapping of trustworthiness to a metric is also possible.

[0018] Within the scope of this disclosure of the invention, an increase in a metric is preferably classified as an improvement in trustworthiness, and a decrease in a metric as a deterioration in trustworthiness. However, the invention is not limited to this. Rather, the person skilled in the art will immediately recognize that the invention can also be realized by a corresponding or equivalent definition of a metric, wherein a decrease in the metric corresponds to an increase in trustworthiness and an increase in the metric to a decrease in trustworthiness. In one embodiment, the higher the trustworthiness, the more likely a telephone call can be forwarded to the recipient, while a low trustworthiness in this embodiment is more likely to result in a call not being forwarded to the recipient, including being forwarded to voicemail.Alternatively, call forwarding is possible regardless of the metric, or at least within a certain range (e.g., above a value greater than or equal to a first threshold), with the metric being transmitted to the recipient for display or other indication. This allows the recipient to decide whether to accept a call or not. A metric thus allows for a numerical assessment of a call's trustworthiness. The value of a metric can, for example, be compared to a threshold to arrive at a technically feasible decision regarding whether to forward or not forward a phone call, or to provide appropriate notification (e.g., visually, using a first color below the threshold and a second color above the threshold).

[0019] The aforementioned interaction metric of the TNI (e.g., TNI A) in relation to a TNI from its interaction list (e.g., TNI B) thus evaluates a direct telephone call relationship. This makes it possible to characterize not only the individual TNI (e.g., TNI A) but also its individual telephone call relationships (e.g., to B and to other TNIs from A's interaction list) in a differentiated manner. For example, the call metric of a telephone call from the originator (e.g., with TNI A) to the recipient (e.g., with TNI B) can exhibit high trustworthiness. This is the case, for instance, if the originator A was previously called by the recipient B. This can be possible even though the interaction set of the originator TNI (A) otherwise contains many interaction metrics with low trustworthiness (for example, because the originator has many other participants with only outgoing calls).Furthermore, the call metric can be influenced by the existence of indirect relationships in the social network graph.

[0020] According to aspects of the invention, the detected telephone call is processed based on the call metrics, wherein the processing includes at least forwarding the telephone call to the recipient or not forwarding the detected telephone call. As explained above, not forwarding can, for example, correspond to rejecting the telephone call, but also to displaying a warning on the recipient's terminal display or forwarding the call to a voicemail box.

[0021] In a variation of the aforementioned embodiment, embodiments of the invention are also conceivable in which the use of an interaction metric is dispensed with, and each participant is assigned an absolute score—instead of a score related to another participant (TNI)—to characterize trustworthiness. For example, the participant's telephone behavior could be reflected in this score. In such an embodiment, determining the call metric for the telephone call could be limited to the fact that the calling participant has a high score, i.e., a score that is above a threshold, and this score resulted from the fact that the calling participant has had past relationships with other participants in the social network graph, and these other participants are also related to the called participant.

[0022] Preferably, the interaction set assigned to the originator and / or the receiver is updated based on at least one parameter of the phone call, with the update preferably including an adjustment of the interaction metric. In some embodiments, the at least one parameter of the phone call can include the receiver's acceptance or rejection of the call or the duration of the call. In simplified terms, the outcome of the phone call thus influences the interaction sets.

[0023] Updating the list of participant identifiers, the originator interaction set assigned to the originator, and the receiver interaction set assigned to the receiver, based on at least one parameter of the phone call, updates the originator interaction set for future phone calls. For example, if the call is accepted or if the call exceeds a certain threshold, a new direct phone call relationship can be created by assigning the receiver to the originator's interaction set. Furthermore, a new direct phone call relationship can be created by assigning the originator to the receiver's interaction set. Additionally, the corresponding interaction metrics in the originator interaction set can be updated based on the call outcome.In some cases, this can affect not only the originator's interaction metric with the call recipient, but also the originator's interaction metrics with other TNIs within their interaction set. Thus, the outcome of the call can impact not only the phone call relationship with the recipient, but also other phone call relationships of the originator. The outcome of the phone call can encompass a range of events, such as the call being actively rejected by the recipient, the call being left unanswered and terminated by the telephony network after a certain time (because "no one answered"), the call being answered and lasting x seconds, or the originator terminating a left-unanswered call after x seconds. The corresponding metrics can be updated based on such events.For example, a metric can be implemented not only based on whether a call was answered, but also, for example, on the duration of a corresponding answered call, or on the duration until an unanswered call was terminated by the originator.

[0024] By considering direct and / or indirect telephone call relationships between participants in a telephony network, the reliability and usability of telephone call monitoring with regard to forwarding or not forwarding detected calls is improved. In particular, this method can be used in a telephony network to reduce call load based on reliable and user-friendly automatic filtering of telephone calls.

[0025] In some embodiments of the invention, if the interaction set of further participant identifiers assigned to the originator TNI does not include an assignment of the receiver TNI, updating the interaction set assigned to the originator and / or the receiver may include including the receiver TNI in the interaction set assigned to the originator TNI and / or including the originator TNI in the interaction set assigned to the receiver TNI.

[0026] In this case, the originator TNI's interaction set does not contain a receiver assignment. It is also possible to use an initial metric for the telephone call relationship between the originator and the receiver. If the originator first appears in the telephony service network with a call to the receiver, the originator TNI can be added to the list of subscriber identifiers, and the receiver TNI can be assigned to the originator TNI's interaction set with the initial interaction metric.

[0027] The initial metric can be set to any value within the range of the metric used. For a newly added participant or a new phone call relationship, the value of the initial metric represents a "trust presumption." Possible values ​​are 0.5, 0.8, or 1.0 (within the aforementioned range of 0.0 to 1.0).

[0028] Determining the call metric for a new phone call relationship is based on the interaction set of the originator TNI and the originator's initial interaction metric with respect to the recipient. If the originator TNI's interaction set already includes other TNIs (e.g., because the originator is already known in the telephony service network and has made previous calls), the call metric can depend not only on the initial interaction metric with respect to the recipient but also on the interaction metrics of the existing TNIs. For example, if the interaction metrics of the other TNIs in the originator's interaction list are predominantly low, the call metric can be significantly reduced compared to the initial metric, resulting in a low call metric being assigned to the originator's call to the recipient.The "probable trust" given with the initial metric is therefore evaluated when determining the call metric based on the originator's call history and their other telephone call relationships.

[0029] If the originator has no other TNIs in the interaction set (e.g., because the originator is appearing in the telephony service network for the first time), the initial metric, and thus the "trust presumption," can be adopted as the call metric. However, a corresponding update of the metric based on the outcome of the phone call, as well as a corresponding update of the metrics based on the outcomes of future phone calls (both to the recipient and other participants), leads to a convergence of the originator's corresponding interaction metrics. Therefore, decisions regarding whether to accept or decline calls from the originator become more reliable with the number of calls made.

[0030] In some embodiments of the invention, updating the interaction set assigned to the originator and / or the receiver may include increasing the originator's interaction metric with respect to the receiver when the telephone call is accepted, the increase being dependent on the duration of the telephone call.

[0031] If a call from the originator is accepted by the recipient, this can increase the originator's trustworthiness within the phone call relationship with the recipient. For example, the value range of the metric can be in the floating-point range (with a decimal point) between 0.0 and 1.0, as described above. Then, the originator's interaction metric with the recipient can be increased by an amount up to the maximum possible value (e.g., 1.0). A smaller increase is also possible, meaning that the originator's trustworthiness within the phone call relationship with the recipient is improved incrementally. For example, the metric can be increased in increments of 0.1. In some implementations, the increase can depend on the current value of the interaction metric. However, the increase can also be small or large, depending on the originator's other interaction metrics.The results may be zero if, for example, a large number of the originator's interaction metrics show a low value. This could indicate that the originator is making mostly unwanted calls, of which only a few are answered by the recipients.

[0032] In some cases, however, the increase may depend on the duration of the accepted phone call. For example, an unwanted call is generally expected to be shorter. Therefore, a longer call duration could indicate a welcome call. For instance, the interaction metric might be increased by an initial value for a call lasting less than 60 seconds and by a second value, greater than the first, for a call lasting longer than 60 seconds.

[0033] In some embodiments of the invention, updating the interaction set assigned to the originator and / or the receiver may include reducing the originator's interaction metric with respect to the receiver if the call was not answered, was rejected by the receiver, or if the call duration was shorter than a call duration threshold.

[0034] In one type of spam call ("ping calls"), the victim's phone rings only very briefly, giving the recipient no chance to answer. The recipient's device then displays a missed call, which can result in higher phone charges if called back. In response to such short calls, the originator's interaction metric with the recipient can be reduced. This reduction can also depend on the duration of the call without the recipient answering, for example, if the originator ends the call after less than two seconds. Furthermore, a call can be considered unwanted if it is deliberately rejected by the recipient—especially if this occurs repeatedly. Finally, the interaction metric for the phone call relationship can be reduced if the call duration was shorter than a predefined call duration threshold.In this case, the call duration threshold can be set to a few seconds (e.g., 15 seconds), corresponding to the time the recipient needs to find out that it is an unwanted call.

[0035] The examples above relate in particular to updating the list of participant identifiers. This means that increasing or decreasing the interaction metrics only takes effect from the next call onwards. In contrast, determining the call metric can already affect the currently detected call. In some embodiments of the invention, updating and / or determining the call metric can therefore include: determining a first number of TNIs in the originator's interaction set that contain at least one received call; determining a second number of TNIs in the originator's interaction set that contain only calls originating from the originator; wherein the originator's interaction metric is increased with respect to the receiver and / or the call metric if the ratio of the first number to the second number is above a call direction threshold.

[0036] Such analyses can include not only interaction metrics for a telephone call relationship but also counters that record the number of incoming and outgoing calls in a telephone call relationship between an originator and a receiver. A TNI used for unsolicited calls (e.g., for telemarketing) will generate a higher number of outgoing calls than incoming calls. However, it is possible that a certain number of incoming calls will also be recorded. A TNI not used for unsolicited calls may have a more balanced ratio of incoming to outgoing calls. Therefore, a TNI that also records a high number of incoming calls can be considered trustworthy.Additionally or alternatively, the call metric can be reduced relative to the originator's interaction metric with respect to the receiver if the ratio of the first to the second count falls below a call direction threshold. In this case, significantly more calls originate from the originator than are received, indicating a connection from which unwanted calls frequently originate.

[0037] In particular, as described above, this method can be used to identify TNIs that may be generating many unwanted calls, whereas TNIs that are not used for such activities tend to have a more balanced ratio of incoming to outgoing calls.

[0038] In some embodiments of the invention, determining the call metric may include determining whether the originator's interaction set contains the receiver's TNI with at least one telephone call received by the originator, wherein the call metric is increased relative to the originator's interaction metric with respect to the receiver if the originator's interaction set contains the receiver's TNI with at least a first number of calls received by the originator, and / or wherein the call metric is decreased relative to the originator's interaction metric with respect to the receiver if the originator's interaction set contains the receiver's TNI with at least a second number of calls made by the originator.

[0039] Such embodiments of the invention relate in particular to the telephone call relationship between a specific originator and a specific recipient. This can be used to counter a form of telephone harassment originating from one person and directed against another (stalking). The call metric is increased if at least an initial number of telephone calls have been made by the recipient to the originator. It is assumed that calls from the originator are welcome to the recipient if the recipient has called the originator several times. Conversely, calls from the originator to the recipient could be unwelcome if the originator makes at least a second number of calls to the recipient. In some embodiments, the reduction can depend on whether the second number of outgoing calls from the originator is (significantly) higher than the first number of incoming calls.

[0040] In some embodiments of the invention, the call metric can be determined by: mapping a relationship between the receiver TNI and the originator TNI in a tree diagram; determining whether the receiver TNI is findable as a result of an N-step search in the tree diagram starting from the originator TNI or the interaction set associated with the originator TNI. The call metric cannot be changed or reduced relative to the originator's interaction metric with respect to the receiver if the number N of steps in the search exceeds a threshold value.

[0041] In some cases, there is no direct telephone call relationship between the originator and the recipient, but rather an indirect one. This occurs, for example, when the recipient's TNI is not part of the originator's interaction set, but is part of the interaction set of a third TNI, which in turn is part of the originator's interaction set. In such cases, both the recipient and the originator of the call are part of a third party's social network. The examples therefore involve determining whether direct or indirect telephone call relationships exist between the originator and the recipient, and if so, with how many levels (nodes).

[0042] Since the originator's interaction set contains TNIs, which in turn are associated with an interaction set, the receiver can be found as the result of an N-stage tree search starting from the originator, based on the interaction set associated with the originator TNI. This tree search can be defined recursively, where the first stage of the tree search corresponds to a search for the receiver TNI in the originator's interaction set, and the Nth stage of the tree search, starting from the second stage, corresponds to a search for the receiver TNI in the interaction set of each TNI in the interaction set of the (N-1)th stage of the tree search.

[0043] However, it should be noted that determining whether the recipient TNI is findable as a result of an N-step tree search starting from the interaction set associated with the originator TNI can also be achieved through other procedural steps besides the steps of a recursive tree search starting from the originator TNI. For example, social networks with direct and indirect telephone call relationships might already be stored as clusters in a device of the telephony service network. Furthermore, indirect telephone call relationships might already be indexed for at least some TNIs in the subscriber identifier list of a telephony service network, so that indirect telephone call relationships for a TNI in the list are directly retrievable.Thus, determining the call metric and the above-mentioned determination can include any steps that produce the result of determining whether the recipient TNI is reachable via N nodes in a social network that records direct and indirect telephone call relationships.

[0044] In some embodiments of the invention, the call metric cannot be changed or decreased relative to the originator's interaction metric with respect to the receiver if the number of levels of the tree search exceeds a threshold value. In some such embodiments, the call metric can initially be increased if, for example, an indirect telephone call relationship exists at a second level between the originator and receiver via exactly one third TNI. From a certain degree or level of an indirect telephone call relationship onward, it can be assumed in this embodiment that a call from the originator to the receiver is to be considered more trustworthy. For example, an indirect telephone call relationship between the originator and the receiver spanning more than two levels might not indicate increased trustworthiness.

[0045] In some cases, the call metric can even be reduced relative to the originator's interaction metric with respect to the receiver, corresponding to the number of levels in the tree search. However, in other cases, the originator's interaction metric may not change with respect to the degree of the indirect phone call relationship if a certain degree (i.e., a threshold of levels) is exceeded. For example, more than two levels might not result in a change to the interaction metric. In this case, the determination of the number of levels in the phone call relationship between the originator and the receiver can be stopped at the aforementioned threshold of levels.

[0046] In some embodiments of the invention, the call metric can be increased relative to the originator's interaction metric with respect to the receiver according to the frequency with which the receiver TNI is found at a given stage of the tree search. In such embodiments, indirect telephone call relationships can exist between the originator and the receiver via multiple third-party participants in the telephony service network. Such indirect multiple relationships can increase the originator's trustworthiness with the receiver.

[0047] In some embodiments of the invention, the call metric can be reduced relative to the originator's interaction metric with respect to the receiver if call volume in the telephony service network exceeds a network load threshold. If call volume, i.e., the number of telephone calls of a certain duration per unit of time (e.g., measured in Erlang), is greater than the number of calls the network can handle, network congestion can occur. In particular, with increasing call volume, the probability of a telephone call being blocked due to congestion also increases. To prevent blocking due to network congestion, the call metrics of calls in the network can be reduced or increased depending on the call volume in the telephony service network.In particular, a call metric can be reduced relative to the originator's interaction metric with respect to the recipient if call volume on the telephony network exceeds a network load threshold. In these cases, an additional proportion of calls that would be forwarded to the recipient under lower network load are not forwarded. This effectively filters calls more strictly based on whether they are desired or unwanted, with calls of high trustworthiness still being forwarded to the recipient, while the additional portion not forwarded consists primarily of unwanted calls. This allows for call prioritization that is acceptable from the user's perspective.

[0048] Aspects of the invention further include a device according to claim 13.

[0049] Aspects of the invention further include a computer program according to claim 14. Detailed description

[0050] The following are some important, non-exhaustive, aspects of the invention explained with the aid of figures. They show: Fig. 1 : a telephony service network according to aspects of the invention; Fig. 2 : a first list of participant identifiers (TNIs) according to aspects of the invention; Fig. 3 : a second list of participant identifiers (TNIs) according to aspects of the invention; Fig. 4 : a third list of participant identifiers (TNIs) according to aspects of the invention; Fig. 5 : a fourth list of participant identifiers (TNIs) according to aspects of the invention; Fig. 6 : a flowchart of the method for monitoring telephone calls according to aspects of the invention.

[0051] Fig. 1Figure 1 shows an exemplary telephony service network according to aspects of the invention. The telephony service network 1000 may be a fixed-line telephony service network, e.g., a PSTN, a telephony service network of a mobile operator, or a combination of both.

[0052] Telephony Service Network 1000 comprises, for example, three users: subscriber A 150, subscriber B 250, and subscriber D 450, whose subscriber lines are registered and managed within Telephony Service Network 1000. In contrast, the subscriber line of subscriber C 350 is registered and managed in a network outside of Telephony Service Network 1000 (e.g., in a network of a different network operator and / or in a different country). The lines of subscribers A 150, B 250, C 350, and D 450 can be either landline or mobile phone lines. Each of these subscribers is identifiable by a unique subscriber identifier (TNI), which could be, for example, a Session Initiation Protocol (SIP) address, a mobile network International Subscriber Identity (IMSI), or another unique subscriber line identifier.

[0053] The telephony service network 1000 also includes a switching device 1200, which switches telephone calls between subscriber terminals in the telephony service network 1000 as well as to and from subscriber terminals outside the telephony service network 1000.

[0054] The switching device 1200 further comprises a telephone call monitoring device 1220. The telephone call monitoring device 1220 is configured to execute the method according to the present disclosure of the invention. For example, the telephone call monitoring device 1220 comprises a processor on which software can be executed, for example, a computer program which includes program instructions which, when the program is executed by a processor, cause the processor to perform all steps of a method according to the invention for monitoring telephone calls according to the present disclosure of the invention. However, it is also possible that the telephone call monitoring device 1220 comprises dedicated hardware which is configured to execute the method according to the present disclosure of the invention.Alternatively or additionally, the monitoring device can include a combination of hardware and software.

[0055] The telephone call monitoring device 1220 can further comprise a list of subscriber identifiers 2000 according to the present invention. The list 2000 can be stored in suitable fixed or volatile memory. The telephone call monitoring device 1220 can read information from the list of subscriber identifiers 2000 and also write, delete, and / or modify information in the list 2000.

[0056] Fig. 2 Figure 1 shows a first list of subscriber identifiers (TNIs) according to aspects of the invention. For example, the list of TNI 2000 is part of the telephone call monitoring device 1220 according to Figure 1. Fig. 1 . Continuing the exemplary example from Fig. 1The TNI 2000 list contains at least entries for users A 150, B 250, C 350, and D 450. The telephone lines of users A 150, B 250, C 350, and D 450 are listed in the 2000 list. Fig. 2 Each is represented by TNI entries 100, 200, 300, and 400. For example, TNIs can include telephone numbers with country code and network or area code. TNIs can be stored in the list regardless of whether they are registered and managed within the Telephony Service Network 1000 or outside of it (e.g., in a network of another network operator and / or another country).

[0057] Each TNI from list 2000 is assigned an Interaction Set (IAM) of subscriber identifiers. In the example, TNI 200 of subscriber B is assigned interaction set 220, TNI 100 of subscriber A is assigned interaction set 120, TNI 300 of subscriber C is assigned interaction set 320, and TNI 400 of subscriber D is assigned interaction set 420. If calls have been made over a telephone line, an interaction set can contain the TNIs of incoming and / or outgoing calls. In this example, the interaction sets of subscribers A, C, and D are not empty, while interaction set 220 of subscriber B is empty. This could be the case if B is a newly configured subscriber who has not yet received or made any calls.In the present example, TNI 100, 300 and 400 of participants A, C and D are each assigned TNI and interaction metrics, which originate from previous calls between the participants.

[0058] For example, interaction set 120 contains an element 123 comprising the TNI of participant C, and an interaction metric of participant A related to participant C, abbreviated as "M(A,C)". An assignment of participant C's TNI, TNI(C), to the interaction set 120 assigned to participant A means that participant A has had incoming and / or outgoing telephone calls with C in the past. In other words, a telephone call relationship exists between participants A and C, or rather, between their TNIs. In the present example, telephone call relationships exist between participants A, C, and D, which are represented in Fig. 2The two-way arrows between TNI 100, 300, and 400, as well as elements 123 and 124 in participant A's interaction set 120, elements 321 and 324 in participant C's interaction set 320, and elements 421 and 423 in participant D's interaction set 420, represent the phone call relationships. The TNI in the list of 2000 participant identifiers and the phone call relationships can be viewed as nodes and edges, respectively, of a social network graph that represents the phone call relationships between participants in a telephony network.

[0059] Although the present case refers to a telephone call relationship, the analysis of relationships according to the invention is not limited to telephone calls. For example, participants A and C could be in contact via social networks (e.g., via messenger services) and exchange messages regularly. In this case, the solution according to the invention could take this into account when calculating the metrics, insofar as the relationship between participants A and C can be mapped to or recognized by the respective TNIs.

[0060] The elements of interaction sets 120, 320, and 420 comprise corresponding metrics M(·,·), each characterizing the telephone call relationships. The value of a metric can characterize a level of trustworthiness, for example, as a floating-point number between 0 and 1, where 1 represents maximum trustworthiness. While the telephone call relationships are symmetrical in nature because they involve both incoming and outgoing calls, the metrics of a telephone call relationship can certainly be asymmetrical. For example, in Fig. 2The metric M(A,C) of participant A with respect to participant C is M(A,C) = 0.94, while the metric M(C,A) of participant C with respect to participant A is M(C,A) = 0.99. This can be due to the fact that, for example, incoming and answered calls for one participant might be assigned different metric values ​​than, for example, incoming and unanswered calls for another. Furthermore, the interaction metrics of telephone call relationships can depend on whether a participant has more outgoing calls than incoming calls. Such asymmetries can be useful, for example, to model stalking, which can be characterized by unilateral, repeated calling of a participant. Here, it is assumed that the difference between M(A,C) = 0.94 and M(C,A) = 0.99 is below a threshold, i.e., small, so that there is a high level of trust on both sides.It can therefore be assumed that calls between participants A and C are not undesirable, regardless of the direction of the call.

[0061] The following considers a call from participant B 250 to participant C. Since participant B's interaction set 220 is empty, the telephone call monitoring device 1220 can assign an initial interaction metric M(B,C) = M o = 0.5 for the first call. If the interaction set is not empty, the telephone call monitoring device 1220 can consider participant B's other interaction metrics in addition to the initial interaction metric to determine the call metric MA of the current telephone call. If predominantly low-value interaction metrics from several other participants X related to participant B, M(B,X), are present, the call metric can be reduced compared to the interaction metric M(B,C). If predominantly high interaction metrics M(B,X) are present, the call metric can be increased compared to the interaction metric M(B,C).

[0062] In the next step, the telephone call monitoring device 1220 can process the call from B to C. For example, a call metric below a lower metric threshold, e.g., 0.2, can prevent the telephone call from being forwarded to subscriber C, without any possibility of reaching the called party. A call metric in the range [0.2 ... 0.5] can result in the caller being forwarded to a voicemail box, which requires verification by voice or keypad input to determine whether, for example, a telephone bot or robocaller is calling. A call metric in the range [0.5 ... 0.8] can result in the caller being forwarded to the voicemail box of subscriber C. For example, the caller (originator) could be asked to briefly state their name, after which the call is immediately forwarded and the recorded message is played back when the call rings.Participant C can then decide whether to accept the call or not based on the played recording. A call metric greater than or equal to an upper metric threshold, e.g., 0.8, can trigger immediate forwarding of the call to the called recipient.

[0063] In this case, it is assumed that the call is not forwarded based on the call metric MA = 0.5, but instead a mailbox is inserted first, requesting the caller to state their name, with the recording then being played back together with the ringtone.

[0064] As a result of the call, the call was accepted by the recipient. Furthermore, the phone call lasted longer than a minimum duration, for example, 923 seconds. Based on this result, interaction set 220 for B and interaction set 320 for C are updated. Instead of a single minimum duration threshold, it is advisable to consider the total duration of the call when adjusting the call metric, for example, according to a direct proportionality.

[0065] Fig. 3 shows a second list of participant identifiers (TNIs) according to aspects of the invention, wherein interaction sets 220 of B and interaction set 320 of C are updated.

[0066] Based on the assumed call and the relatively long call duration of 923 seconds, elements 223 and 322 are added to interaction sets 220 and 320, respectively. The new element 223 in interaction set 220 for participant B contains participant C's TNI, TNI(C), and an increased interaction metric M(B,C) = 0.60. Furthermore, the new element 322 in interaction set 320 for participant C contains participant B's TNI and the interaction metric M(C,B) = 0.6. These new entries in the interaction sets represent a new telephone call relationship between B and C.

[0067] In a later phone call, participant B is called by D. Since participant B's TNI is not part of interaction set 420, which is assigned to TNI 400 in list 2000, an initial assignment with an interaction metric M(D,B) = M o = 0.5 is assumed. However, it can be investigated whether an indirect phone call relationship exists between D and B (where, as already mentioned, the indirect relationship need not be limited to phone calls, but other types of relationships in the network can also be identified). Here, the TNIs of participant D's interaction set 420 can be examined. Interaction set 420 contains an entry 421, which refers to participant A's TNI. However, participant A's interaction set 120 does not contain participant B. Interaction set 420 does, however, contain another entry 423 for participant C, which has a relatively high interaction metric M(D,C) = 0.98.Therefore, the identified indirect telephone call relationship between D and B – via participant C – leads to an increase in the call metric compared to the (initial) interaction metric M(D,B) = 0.5 to MA = 0.8, which is why the call is immediately forwarded to participant B. The call metric can thus depend not only on the interaction metric of the caller (originator) themselves, but also on the interaction metrics in identified direct or indirect telephone call relationships. In this case, participants C and D are already part of a social network, with C having a relatively high interaction metric M(C,B) with the recipient B of the current call (due to the long previous conversation between B and C), which results in a "bonus" for the call metric. It is also possible that falling below a certain interaction metric threshold in an indirect telephone call relationship leads to a decrease in the call metric.For example, the previous phone call between B and C might have been rejected, resulting in an interaction metric M(C,B)=0.35. Since this is below a threshold of, for example, 0.4, the indirect phone call relationship between D and B is evaluated such that the call metric is not increased compared to the (initial) interaction metric M(D,B)=0.5.

[0068] In the present example, B accepts D's call, and the call duration is 142 seconds. Since this is below a call duration threshold of 150 seconds, the updated interaction metric in interaction sets 420 and 220 for participants D and B is assigned the value M(D,B)=M(B,D)=0.8. Fig. 4The new elements 422 and 224 are shown in the interaction sets. Furthermore, it should be noted that the social network has another edge between D and B, which symbolizes a direct telephone call relationship.

[0069] In a subsequent phone call, participant B calls participant A. Participant A has configured the telephone call monitoring device 1220 to not forward calls below a call metric of 0.9 (i.e., from likely unknown phone numbers). Since participant B's interaction set 220 does not yet contain an assignment to the TNI(A) (see below), the interaction set 220 is not yet configured. Fig. 4Initially, an assignment with an initial interaction metric M(B,A) = M o = 0.5 is assumed. Furthermore, an indirect telephone call relationship between B and A is established, both via participant C (through elements 223 and 321 of participant B's interaction sets 220 and 320 of participant C) and via participant D (through elements 224 and 421 of participant B's interaction sets 220 and 420 of participant D). Due to the multiple indirect telephone call relationships between B and A, the call metric is increased from the initial metric M(B,A) = M o = 0.5 to MA = 0.9. Based on this call metric, the call is forwarded to A without further ado. The call received by A lasts, for example, 1024 seconds.

[0070] Fig. 5This shows an example of a fourth list of participant identifiers (TNIs), which results from the long phone call between A and B. The list contains the new elements 221 and 122, in which the call metrics M(B,A) and M(A,B) have been increased to the value 0.95, respectively. This is due, on the one hand, to the long duration of the call, and on the other hand, because of the double phone call relationship, which takes into account that A was already at least indirectly part of a social network of several participants before the call, all of whom had already been in contact with A. For this reason, the interaction sets 320 and 420 of participants C and D with respect to participant B in elements 322 and 422 are also increased to the value 0.95. The step of updating the list of participant identifiers based on the outcome of the phone call can therefore not only be directly attributed to the originator or...not only affect the recipient of the telephone call, but also the interaction sets regarding TNI, through which an indirect telephone call relationship exists.

[0071] Fig. 6Figure 1 shows a flowchart of the method for monitoring telephone calls according to aspects of the invention. In step 10, it is determined that a telephone call is made from an originator to a receiver, wherein the originator and the receiver are each assigned a subscriber identifier (TNI) in a list of subscriber identifiers. In particular, each TNI from the list of subscriber identifiers is assigned an interaction set of further subscriber identifiers, and each of these assignments includes an interaction metric based on the TNI and the respective further TNI in the interaction set. In step 20, a call metric for the telephone call is determined based on the interaction set of the originator TNI and the interaction set of the receiver TNI.In step 30, the telephone call is processed based on the call metric, wherein the processing includes at least forwarding the telephone call to the recipient or not forwarding it. Preferably, in step 40, the interaction set assigned to the originator and / or the recipient is updated based on at least one parameter of the telephone call, wherein the update preferably includes an adjustment of the interaction metric, as already explained in detail above.

[0072] In the present disclosure of the invention, exemplary rules and values ​​have been mentioned according to which a call metric is determined depending on certain conditions, such as the existence of direct and indirect telephone call relationships or other social network relationships between an originator and a receiver of a telephone call. Furthermore, exemplary rules and values ​​have been mentioned according to which the interaction metrics relating to subscriber identifiers in a telephony service network can be updated based on the outcome of a telephone call between an originator and a receiver. It should be noted that the individual rules and values ​​mentioned are not exhaustive. Rather, the determination of a call metric and the updated interaction metrics can be based on the interaction of several rules, including further rules and values ​​not included in the present disclosure of the invention.In one embodiment, the rules and values ​​are based on procedures that are trained and determined using machine learning (ML) algorithms.

Claims

1. Method for monitoring telephone calls, the method comprising the following steps: Determining (10) that a telephone call is being made from an originator to a recipient, wherein the originator and the recipient are each associated with a subscriber identifier, TNI, in a list of subscriber identifiers (2000), wherein each TNI (100, 200, 300, 400) from the list of subscriber identifiers is associated with an interaction set of further subscriber identifiers (120, 220, 320, 420), and wherein each of these associations comprises an interaction metric (123, 124; 321, 324; 421, 423) with respect to the TNI and the respective further TNI of the interaction set; Determining (20) a call metric for the telephone call, based on the interaction set of the originator TNI and the interaction set of the recipient TNI; Processing (30) the telephone call based on the call metric, wherein the processing comprises at least one of forwarding to the recipient or not forwarding the telephone call, wherein the call metric is reduced relative to the interaction metric of the originator with respect to the recipient, if a call volume in the telephony service network exceeds a network load threshold.

2. Method according to claim 1, further comprising: Updating (40) the interaction set associated with the originator and / or the recipient based on at least one parameter of the telephone call, wherein the updating preferably comprises an adjustment of the interaction metric.

3. Method according to claim 2, wherein the at least one parameter of the telephone call comprises the acceptance or non-acceptance of the telephone call by the recipient, or the duration of the conversation.

4. Method according to any one of claims 2 or 3, wherein the interaction set (220) of further subscriber identifiers associated with the originator TNI does not comprise an association of the recipient TNI, and the updating (40) of the interaction set associated with the originator and / or the recipient comprises an inclusion of the recipient TNI in the interaction set associated with the originator TNI and / or an inclusion of the originator TNI in the interaction set associated with the recipient TNI.

5. Method according to any one of claims 2 to 4, wherein the updating (40) of the interaction set associated with the originator and / or the recipient comprises increasing the interaction metric of the originator with respect to the recipient, if the telephone call was accepted.

6. Method according to claim 5, wherein the increase depends on the duration of the telephone call.

7. Method according to any one of claims 2 to 4, wherein the updating (40) of the interaction set associated with the originator and / or the recipient comprises decreasing the interaction metric of the originator with respect to the recipient, if the call was not accepted, was rejected by the recipient, or if the duration of the call was shorter than a call duration threshold.

8. Method according to any one of claims 2 to 7, wherein the updating (40) of the interaction set associated with the originator and / or the recipient comprises: Determining a first number of TNIs in the interaction set of the originator that have received at least one incoming call; Determining a second number of TNIs in the interaction set of the originator that have only outgoing calls from the originator; wherein the interaction metric of the originator with respect to the recipient is increased if the ratio of the first number to the second number is above a call direction threshold.

9. Method according to any one of the preceding claims, wherein determining the call metric comprises: Determining a first number of TNIs in the interaction set of the originator that have received at least one incoming call; Determining a second number of TNIs in the interaction set of the originator that have only outgoing calls from the originator; wherein the call metric is increased relative to the interaction metric of the originator with respect to the recipient, if the ratio of the first number to the second number is above a call direction threshold.

10. Method according to any one of the preceding claims, wherein determining the call metric is obtainable by: Mapping a relationship between the recipient TNI and the originator TNI in a tree diagram; Determining whether the recipient TNI is obtainable as a result of an N-level search in the tree diagram, starting from the originator TNI or the interaction set associated with the originator TNI.

11. Method according to claim 10, wherein the call metric is not changed or is reduced relative to the interaction metric of the originator with respect to the recipient, if the number N of the search corresponds to more than a threshold value of levels.

12. Method according to claim 10, wherein the call metric is increased relative to the interaction metric of the originator with respect to the recipient, in accordance with a frequency indicating how often the recipient TNI is obtainable in a level of the search.

13. Device comprising means configured to carry out the method according to any one of claims 1 to 12.

14. Computer program comprising program instructions which, when the program is executed by a computer, cause the computer to carry out all steps of the method according to any one of claims 1 to 12.