Background Networked processing systems can nowadays be used for a variety of data processing applications, for example analysing and extracting available statistics from third-party sources for use with a target data processing application.

Such applications include the gathering of statistics from users to compute factors such as life expectancy, or the probabilistic expectation of a certain event such as an illness, injury, personal accident or vehicle accident, based upon which an insurance estimate may be given. However, there are technical challenges in implementing a system which can gather the statistics over a network in an efficient and secure manner. Traditionally, users are required to contact multiple insurance providers in order to obtain and compare estimations for their insurance products. The users may call a customer service number or fill in an online form for each of the insurance providers to provide personal details and other relevant information. For example, the users may provide their occupation and a list of existing health conditions when requesting an estimate for life or health insurance, or they may need to submit vehicle details and a history of accidents when obtaining estimates for automotive insurance. Such tasks are time consuming and repetitive, and thus limit the number of insurance providers a potential customer may contact when requesting insurance estimates.

Insurance comparison sites such as comparethemarket.com provide an efficient way for users to enter their details for use with one or more insurance providers. More specifically, the users may fill in a single form at an insurance comparison site so that their details are distributed to a selection of insurance providers. The details entered in the single form comprise textual and/or numerical data that are typically required in obtaining insurance estimates, as well as the levels of insurance cover required. In response, some or all of the insurance providers each returns an estimate based on the details submitted to them, and thereupon the insurance comparison site compiles and displays a list of returned estimates to the user. Summary

The prior art method described above attempts to improve the efficacy of obtaining insurance estimates from multiple insurance providers. Not only does it require less repetitive data entry by the customers at each individual insurance provider's website, but the insurance providers previously unknown to the customers may also provide their estimates for said customers' consideration. Nevertheless, the users are still required to fill in a complete form at the insurance comparison website to provide the necessary details, and therefore it still takes a degree of effort from the user. Moreover, the physical entry of personal data in a public place, e.g. at an internet cafe, may make the user an easy target for prying eyes.

Furthermore, details provided to each of the insurance providers are not error-proof, and their accuracy depends on what was manually inputted into the single form at the insurance comparison site. As such, some inconsistencies may occur unintentionally, e.g. typographic errors or if the user has misunderstood the question in the form. Furthermore, it is not uncommon that users may provide fraudulent information, e.g. date of birth, lifestyle, accident history, in order to lower their insurance premium.

In addition, the users' details are passed onto each of the insurance providers for evaluation at their end. This exposes security weaknesses in that the users' details may be stored at each of the insurance providers' databases for future unauthorised use, e.g. marketing, or being illicitly distributed to third parties, or being hacked by malicious parties. The users' details are also susceptible to being intercepted when transmitting from the insurance comparison website to each of the insurance provider's server. When the details are distributed to multiple providers, this multiplies the number of points of susceptibility.

The present disclosure provides a system that allows the users' attributes to be extracted from their social media profiles, and based on analysing said users' attributes compute one or more analytics outputs for determine a risk of the users. The system may, with the use of one or more analytics outputs, either estimate an insurance premium thereat, or it may share the one or more analytics outputs to the insurance providers so that the estimation of insurance premium can be carried out at their end. This reduces, or even eliminates, the need for users to fill in a form for estimating insurance premiums. Advantageously, the attributes gathered from the users' social media profiles are more secure as they are not vulnerable to key logger attacks or even a malicious party watching over the user's shoulder as he /she types (at least not at the point of requesting the insurance estimate). It may also help preventing fraud because the users' social media profile may have already been subject to public scrutiny. In some embodiments, a picture of the user may be captured from said user's social media profile or by a user terminal (e.g. a smartphone), in order to determine biometric information or attributes of the user, e.g. age, gender and/or race, and/or to reveal the user's lifestyle, e.g. are they heavy smokers or alcoholics.

In some embodiments insurance premiums are estimated centrally using estimation algorithms provided to the central point by each of the insurance providers, such that the users' details are not required to be distributed to third parties, let alone multiple parties. More specifically, the users' details are kept at a service provider's system where estimation is carried out locally based on the algorithms gathered from the third parties. This enhances security by eliminating any risks associated with data transmission and trust issues with third parties providers.

According to a first aspect disclosed herein there is provided a server, comprising: an interface for connecting to one or more social media systems, and to one or more insurance provider databases of a respective one or more insurance providers; memory comprising one or more memory units, said memory storing code; processing apparatus comprising one or more processing units, said processing apparatus being arranged to run said code; wherein the code is configured so as when run on the processing apparatus to perform operations of: accessing a user profile of a user from each of one or more social media systems, and extracting therefrom one or more attributes of the user; at the server, analysing the one or more attributes of the user to compute one or more analytics outputs indicative of a risk and/or one or more insurance criteria associated with the user; and providing the one or more analytical outputs to a respective estimation algorithm of each of the one or more insurance providers to produce a respective insurance estimate from each of the one or more insurance providers.

Thus the present disclosure provides a method and system for gathering one or more attributes the users from the users' social media profiles. Said attributes, such as age and gender, may be extracted from the users' social media profiles on one or more social media systems, e.g. Facebook and/or Linkedin, before being analysed at the system to compute one or more analytics outputs. For example, the attributes are combined to from one or more analytics outputs. More specifically, said analytics outputs require further processing of the extracted attributes and are not immediately apparent in the attributes as extracted from the user profiles. The one or more analytics output may comprise quality of social circle and/or family members, an inferred relationship status, health condition, activity engaged in by the user, travel history, income range and/or risk presented to the user.

For example, whilst users' attributes such as age and gender can be readily extracted from said users' profiles via a screen scraping technique, analytics output such as the users' health condition would require further analysis based on the users' age and gender. The one or more analytics outputs are indicative of a risk associated with the user, and are used in estimation insurance estimates, or insurance premium, of the user.

In some embodiments, age, gender and at least one additional attribute of the user are extracted from the user profile. In some embodiments, the one or more attributes of the user comprise at least one attribute other than age and gender. In some embodiments, the one or more analytics outputs are indicative of a risk of user falling ill, having an accident and/or dying.

The system and method may be applicable to any insurance product, for example life insurance, health insurance, automotive insurance and home insurance. Further, the social media systems are not necessary limited to the traditional social media systems, but more generally they may be any online service provider having a database storing user details, such as electoral register, employee database, online auction websites, etc.

In embodiments, the code may be configured so as when run on the processing apparatus to perform the further operations of: receiving said respective estimation algorithm from each of the one or more insurance providers; at the server, estimating said respective insurance estimate based on the respective estimation algorithm from each respective one of the one or more insurance providers, wherein each of the estimates is based on the respective estimation algorithm of the respective insurance provider applied to said one or more analytics outputs; and providing a list comprising the estimates of each of the one or more insurance providers to the user.

In embodiments, the code may be configured so as when run on the processing apparatus to perform the further operations of: estimating an insurance estimate based on said estimation algorithm applied to said one or more analytics outputs; and providing the estimate to the user.

In embodiments, the code may be configured so as when run on the processing apparatus to perform the further operations of: sending the one or more analytics outputs to one or more insurance providers, and receiving said respective estimate from each respective one of the one or more insurance provider, wherein each of the estimate is based on the one or more analytics outputs. In embodiments, the code may be configured to receive the user profile by screen scraping from a respective webpage of each of the one or more social media systems.

In embodiments, the code may be configured to receive the user profile by extracting user profiles from a respective API of each of the one or more social media systems.

In embodiments, the one or more attributes of the user may comprise at least one attribute other than age and gender.

In embodiments, the one or more attributes may comprise age, gender and at least one further attribute.

In embodiments the one or more attributes may comprise one or more of: occupation, education, place of residence and/or contacts of the user. In embodiments at least one of the one or more analytical outputs may be computed by combing two or more of said attributes.

In embodiments, the one or more one or more analytics outputs may comprise one or more of: a metric representative of a social circle and/or family of the user, an inferred relationship status, a health condition, an activity engaged in by the user, travel history and/or income range.

In embodiments, the metric representing the user's social circle and/or family may comprise a number of contacts of the user and/or a frequency of interaction with the contacts of the user. In embodiments, the code may be further configured to extract identities of one or more contacts of the user from the user profile, and based thereon to extract one or more attributes from one or more social media profiles of the contacts; and wherein the metric representing the user's social circle and/or family is generated based on the one or more extracted attributes of the one or more social media profiles of the one or more contacts.

In embodiments, the one or more analytics outputs may comprise at least one metric indicative of a risk of the user falling ill, having an accident and/or dying.

In embodiments the one or more attributes of the user may comprise one or more images, and the code may be configured to generate at least one of the one or more analytics outputs based on said one or more images.

In embodiments, at least one of the images may be retrieved from one of the one or more social media systems.

In embodiments, at least one of the images may be captured by a camera.

In embodiments, the code may be configured to verify the image captured by the camera by comparing with the image retrieved from social media or vice versa.

In embodiments, the code may be configured to use the at least one analytics output generated from the one or more images to verify at least one other of the attributes of analytics outputs.

In embodiments the verified attribute or analytical output may comprises an age of the user.

In embodiments, at least one of the images may comprise facial features of the user, and the code may be configured to generate said at least one analytics output at least in part based on one or more of the facial features.

In embodiments, said one or more facial features may comprise one or more of: gender, skin condition, dental condition, sightedness, hair colour and/or hair loss of the user. In embodiments, at least one of the images may comprise one or more bodily features of the user, and the code may be configured to generate said at least one analytics output at least in part based on one or more of the bodily features.

In embodiments, said one or more bodily features may comprise one or more of: size, proportion, and/or one or more physical impairments of the user.

In embodiments, the code may be further configured to identify one or more background elements in at least one of the images, and to infer at least one of the analytics outputs from the one or more background elements.

In embodiments, the one or more inferred analytics outputs may comprise one or more of: a hobby, occupation and/or other activity engaged in by the user.

In embodiments, the estimation algorithm may be configured to verify one or more of the analytics outputs of the user profile based on one or more of the analytics outputs evaluated from the image of the user.

According to a second aspect disclosed herein there is provided a server, comprising: an interface for connecting to one or more social media systems, and to one or more insurance provider databases of a respective one or more insurance providers; memory comprising one or more memory units, said memory storing code; processing apparatus comprising one or more processing units, said processing apparatus being arranged to run said code; wherein the code is configured so as when run on the processing apparatus to perform operations of: accessing a user profile of a user from each of one or more social media systems, and extracting therefrom one or more attributes of the user; providing the one or more attributes, or information derived therefrom, to a respective estimation algorithm of each of the one or more insurance providers to produce a respective insurance estimate from each of the one or more insurance providers. The attributes used preferably comprise at least one attribute other than just age and gender. In embodiments the attributes used may comprise age, gender and at least one further attribute. Accordingly, based on the users' attributes extracted from the user's social media profiles, the system may estimate an insurance premium thereat, or it may share the users' attributes with different insurance providers so that the estimation of insurance premium can be carried out at their end.

According to a third aspect disclosed herein, there is provided a server comprising: an interface for connecting to one or more social media systems, and to one or more insurance provider databases of one or more respective insurance provider databases; memory comprising one or more memory units, said memory storing code; processing apparatus comprising one or more processing units, said processing apparatus being arranged to run said code; wherein the code is configured so as when run on the processing apparatus to perform operations of: accessing a user profile of said user from each of one or more social media systems and extracting one or more attributes of the user; receiving a respective estimation algorithm from each of the one or more insurance providers; at said server, estimating a respective insurance estimate based on the respective estimation algorithm from each respective one of the one or more insurance providers, wherein each of the estimates is based on the respective estimation algorithm of the respective insurance provider applied to one or more of the extracted attributes from the user profile. In embodiments, either of the second or third aspects may comprises any of the features or embodiments associated with the first aspect.

In further aspects there are provides methods of perming operations in accordance with any of the firsts second or third aspects above. In embodiments the method may further comprise steps in accordance with any embodiments of the system disclosed herein.

Brief Description of the Drawings

To assist understanding of the present disclosure and to show how embodiments may be put into effect, reference is made by way of example to the accompanying drawings in which:

Figure 1 is a schematic illustration of the system according to an embodiment of the present invention, Figure 2 is a block diagram of the analytics server according to the embodiment illustrated in Figure 1 ,

Figure 3a is a process flow chart for providing insurance estimates according to an embodiment of the present disclosure,

Figure 3b is a process flow chart for providing insurance estimates according to another embodiment of the present disclosure, Figure 4 is a process flow chart for providing insurance estimates according to yet another embodiment of the present disclosure,

Figure 5 is a process flow chart for providing insurance estimates according to yet another embodiment of the present disclosure,

Figure 6a,b,c,d are screen captures of requesting authorisation and extracting aspects of user profile from social media systems and thereby using said extracted aspects to fill in a form at an interface, and Figure 7 is an image of a user.

Detailed Description of Embodiments

The present invention comprises a system to acquire a user's information, or user's attributes, from one or more social media websites or other such social media systems, and to subsequently compute one or more analytics outputs from analysing said basic information. Said one or more analytics outputs are metrics indicative of a risk associated with the user (e.g. the risk they will fall ill, die or have an accident), derived based on attributes of the user that contribute (e.g. age, weight, etc.). Alternatively or additionally, the one or more analytics outputs may comprise an indication of the insurance needs of the user, i.e. one or more criteria for the product to be provided to the user. Once derived, the analytics outputs are the applied in an estimation algorithm for producing an insurance estimate. In some cases, a list of respective insurance estimates may be generated based on said one or more analytics outputs and an estimation algorithm from each of the respective insurance providers. The user's details may be obtained from any one or more of a variety of sources, such as profile text and/or numerical data in the user's profile information, information concerning the contacts of the user, and/or visual media data such as photos or videos obtained from social media websites or from a camera on a user terminal.

Figure 1 schematically illustrates a system 100 in accordance with embodiments disclosed herein. The system comprises an analytics server 1 10 being configured to connect to a computer network 102 for communicating with a user terminal 122 accessible by a respective user 120, and also with one or more provider servers 130a,b,c each maintained by a respective insurance provider, and with one or more social media systems 140. In the illustrated examples a single user 120 with a single user terminal is featured, but it will be appreciated that there may be different numbers of users 120 each having one or more user terminals 122.

In some cases one or more provider servers 130a,b,c may be a single provider server 130 maintained by a single insurance provider, with only one insurance provider being in communication with the analytics server 1 10 to provide one or more estimation algorithm. Alternatively the one or more provider servers 130a,b,c may be a plurality of provider severs, each operated by a respective insurance provider and each providing a respective one or more estimation algorithms to the analytics server 1 10. Either way, the one or more algorithms provided by a given provider enable the analytics server 1 10 to return one or more estimates for a user. This may be the only one estimate, or a plurality of estimates each detailing the cost for a different product from the given insurance provider. In some cases, the estimation algorithms may be received from the one or more provider servers 130a,b,c shown in Figure 1 via a database (not shown), which retrieves and stores estimation algorithms from each of the insurance providers.

Therefore the system is applicable in the following scenarios: (a) one user requesting insurance estimate from one insurance provider, (b) one user requesting insurance estimates from a plurality of insurance providers, (c) a plurality of users requesting insurance estimates from one insurance provider, and (d) a plurality of users requesting insurance estimates from a plurality of insurance providers.

In some other embodiments, the one or more algorithms are not provided to the analytics server 1 10. Instead, the analytics outputs of a user are sent from the analytics server 1 10 to each of the one or more providers servers 130a,b,c for estimating insurance estimates thereat. More specifically, the one or more provider servers 130a,b,c are each configured to store its own estimation algorithm, such that based on the received analytics outputs they estimate insurance estimates to be returned to the analytics server 100.

In yet other embodiments, the analytics server 110 is in fact a server that does not require frequent interaction with a provider server 130a,b,c, or in some case the analytics server 100 does not connect to any provider server at all. Similar to the other cases, the analytics server 1 10 analyses extracted user's attributes from the user's profiles to compute one or more user's one or more analytics outputs. However, the analytics server 1 10 in this case estimates premiums based on a set of standard estimation algorithms. Such a standard estimation algorithm may be a predetermined estimation algorithm or a composite estimation algorithm produced by combining estimation algorithms from multiple insurance providers. The objective of an analytics server 1 10 is to provide user a ballpark estimate that is independent of insurance provider.

Thus the estimation of insurance premium is carried out in one of the follow ways: (i) the analytics server 1 10 estimates the premium centrally using estimation algorithms provided by the insurance providers, (ii) the insurance providers estimate premiums at their ends using the one or more analytics outputs provided by the analytics server 1 10, or (iii) the analytics server 1 10 estimates the premium centrally using a standard estimation algorithm.

Note that a server as referred to herein may comprise one or more physical server units located at one or more geographic sites (distributed storage and processing techniques in themselves being known in the art).

The network 102 is preferably a packet-based network, and preferably a public network. In embodiments it may take the form of a wide-area internetwork such as that commonly referred to as the Internet. The user terminal 122 may be connected to the main network 102 via a client side network 104, for example, a mobile cellular network, or a wireless local area network (WLAN), or a combination of any of these.

The analytics server 1 10 is connected to the network (e.g. Internet) 102, and is arranged to serve the user terminal 122 via the network 102, and to gather the user's attributes from the social media systems 140 and provider servers 130a,b,c via the network 102. The analytics server 1 10 is tasked with the following main functions: (i) to retrieve the user's attributes from a user's profile from the social media systems 140, and/or to retrieve a user's image or video from the user terminal 122 or from one of the social media systems 140; (ii) to analyse the extracted the user's attributes and/or retrieved images or video to output one or more analytics outputs; and (iii) based on said analytics outputs either estimate an insurance premium centrally or dispatch the analytics outputs to the insurance providers for estimating premium thereat.

The user terminal 122 may take any of a variety of different forms, such as a desktop computer, laptop, tablet, smartphone, smart watch, pair of smart glasses, smart TV, set-top box, or a digital camera. Where there is more than one user terminal 122 present, the different user terminals 122 need not necessarily take the same form as one another. Note therefore that the term "computer" as used herein does not restrict to a traditional desktop or laptop computer. In some embodiments, the user terminal 122 comprises an on-board camera 124 for capturing an image (or video) of the user. The camera 124 may be positioned by the screen at the front of the user terminal 122, such that a "selfie" image of the user 120 can be taken. Or else the camera can be positioned at the rear surface of the user terminal 122 for capturing an image of said user, or indeed can be positioned elsewhere (e.g. peripheral webcam).

A client application is installed on computer-readable storage of its respective user terminal 122, and arranged for execution on a respective processing apparatus of the respective user terminal 122. The storage may take the form of one or more storage media (e.g. magnetic memory, electronic memory and/or optical memory) implemented in one or more memory units. The processing apparatus 1 12 may take the form of one or more processing units. Each of the client applications is configured so as to be able to communicate with the analytics server 1 10 via the network 102 (and in embodiments via the respective local network) in order to access the insurance estimates. The client 122 may for example take the form of a web browser configured to access the estimates via a website hosted by the analytics server 1 10. As another example, the client 122 may take the form of a dedicated client app provided by the operator of the service hosted by the analytics server 1 10.

Figure 2 is a block diagram illustrating the main components in the analytics server 1 10. As shown in Figure 2, the analytics server 1 10 comprises a network interface 1 1 1 , user profile database 114, algorithm database 116 and estimates database 118 in connection with a processing apparatus 1 12. The network interface 1 1 1 provides an interface for the analytics server 1 10 to connect with the user terminal 122, the one or more social media systems 140 and each of the plurality of provider servers 130a,b,c via the packet based network 102. The processing apparatus 122 is programmed by means of software hosted in memory of the analytics server to perform operation in accordance with the various embodiments disclosed herein.

A process flow chart is provided in Figure 3a to illustrate the steps in performing a first embodiment of the present invention. Starting at step 410, the processing apparatus 112 first receives estimation algorithms from a list of insurance providers. At step 420, it then analyses said estimation algorithms in order to determine the required analytics outputs (e.g. hobbies, etc.). Based thereon, at step 430 the processing apparatus 112 downloads the user's attributes in the user's profile from one or more of the social media systems. At step 440, the processing apparatus 112 then evaluates the user's attributes, and therefrom outputs said required analytics outputs. At step 450 the processing apparatus 112 applies each of the received estimation algorithms to said analytics outputs in order to estimate insurance premium for each of the respective insurance provider. Finally at step 460, the processing apparatus 1 12 communicates the insurance estimates for each of the insurance providers to the user in a ranked list.

An alternative embodiment is shown in Figure 3b, wherein the one or more analytics outputs are extracted from analysing one or more photos or videos of a user. At step 432 the processing apparatus 1 12 receives one or more photos from the user terminal 122 and/or from one or more of the social media systems 140. The photo(s) may be captured by an on-board camera on the user's user terminal 122, or may be stored photo(s) downloaded from the one or more social media systems 140. At step 442, the processing apparatus 1 12 then applies one or more image recognition algorithms 442 to the photo(s) in order to either evaluate attributes of the user (e.g. age, etc) or and to compute one or more analytics outputs (e.g. hobby, etc.). At step 450 the processing apparatus 1 12 applies the estimation algorithms to these extracted aspects in order to estimate insurance premium for each of the respective insurance provider.

In some embodiments, the analytics outputs may be evaluated based on a combination of attributes of the user in the user's profile (steps 430 and 440) and attributes evaluated from photos and videos of the user (steps 432 and 442). This way, missing attributes of the user in a user's profile may otherwise be filled in by attributes obtained from analysing the user's photos and/or videos.

The algorithm database 1 16 stores the list of insurance providers along with their respective estimation algorithms, wherein said estimation algorithms enable the insurance premiums to be estimated based on one or more analytics outputs produced by analysing the attributes of the user in the user profiles. The estimation algorithm may be updated regularly, e.g. updated from the provider servers 130a,b,c on a daily or weekly basis, or the estimation algorithm can be accessed in real time from the provider servers 130a,b,c so to provide the most up to date estimation for the user 120. Based on the estimation algorithms collectively provided by the insurance providers, the processing apparatus 1 12 may compile a list detailing the one or more analytics outputs required for producing estimates for each of the insurance providers.

The analytics outputs are derived based on a plurality of input attributes, i.e. items of input data extracted from the user's profile(s). Some examples of the user's attributes include age, gender, title, occupation, place of residence and/or level of education. The analytics outputs on the other hand are derived by performing analytics on such factors, and represent metrics indicative of a risk that the user will claim on a policy, and/or one or more parameters of the policy to be provided to the user. For example, a user 120's "age" and "occupation" (attributes of said user) may be extracted from said user's user profile from one or more of the social media systems 140, and then an analytics output "health condition" may be outputted and applied in any estimation algorithm that requires the knowledge of the health condition of said user.

In an alternative embodiment as shown in Figure 4, the insurance premium is not centrally estimated by the analytics server 110, but by each of the insurance providers themselves at their respective provider servers 130a,b,c. Similar to the system described in Figure 3a and Figure 3b, the analytics server 1 10 outputs one or more analytics outputs of the user by analysing attributes of the user in the user's social media profile (step 530 and 540) and/or the user's photos and/or videos (step 532 and 542). In step 550, the analytics server 1 10 sends the one or more analytics outputs to the one or more insurance providers so as to enable the insurance premium to be estimated, using a respective estimation algorithm stored thereat. In contrast to an ordinary quotation process where a user provides his/her personal information (i.e. attributes) directly at the insurance provider's website (or an insurance comparison website), this method instead collects and analyses attributes of the user from the user's social media profile, in order to output analytics outputs to the insurance providers. In step 552, the analytics server 1 10 receives one or more insurance quotes from each of the insurance provider, where thereafter in step 560 the processing apparatus 1 12 communicates the insurance estimates for each of the insurance providers to the user in a ranked list.

In an alternative embodiment as shown in Figure 5, the insurance premium is centrally estimated by the analytics server 110 based on a set of one or more standard estimation algorithms. Similar to the system described in Figure 3a and Figure 3b, the analytics server 1 10 output one or more analytics outputs by analysing attributes of the user in the user's social media profile (step 534 and 544) and/or the user's photos and/or videos (step 536 and 546). In step 554 the processing apparatus 1 12 applies a standard estimation algorithm to said one or more analytics outputs in order to estimate insurance premium for the user. The standard estimation algorithm may be a predetermined set of reference estimation algorithms, or it is a composite estimation algorithms averaging from a plurality of estimations algorithms each provided by a discrete insurance provider. Either way, the insurance premium estimated using this method aims to provide the user with a rough estimate over a range of insurance products, e.g. the cost of different levels of insurance coverage.

The processing apparatus 1 12 is programmed with a user profile extractor for extracting a complete user profile or different attributes of the user in the user profile from the one or more social media systems 140, and subsequently storing said complete user profile, e.g. as data files, or attributes of the user in the user profile, e.g. as elements of data files, at the user profile database 1 14. In some embodiments, the user profile extractor may extract user profiles from one or more social media systems each relating to different attributes of the user in user profiles, for example, the user profile extractor may communicate with Facebook for attributes relating to the social information of the user and with Linkedln for attributes relating to the professional information or attributes of said user.

In some embodiments, the processing apparatus 1 12 is configured to access the social media systems 104, in order to retrieve a complete user profile to be stored at the profile database 1 14. The profiles stored at the user profile database 1 14 may be used for further analysis so that different analytics outputs can be extracted therefrom. Advantageously, this enables the processing apparatus 1 12 to carry out further analysis if new analytics outputs are added by insurance providers. Alternatively in some embodiments, the processing apparatus 1 12 may analyse the user's profiles directly at the social media systems 140 and therefrom extract and download only selected attributes of the user in the user's profile. In other words the processing apparatus 1 12 only downloads the one or more attributes of the user in the user profile without downloading the user profile as a whole.

In some embodiments, the processing apparatus 1 12 may extract attributes of the user in the user profiles from a social media system 140 by a screen scrape or web scraping technique. For example, the processing apparatus 1 12 may, upon the user having logged into his/her respective user accounts at the social media systems 140, download displayed attributes of the user in his/her respective user profiles. More specifically, the user profile extractor is configured to analyse onscreen content and extract relevant attributes of the user from the user profiles. Alternatively or additionally, in some embodiments, the processing apparatus 1 12 may extract one or more attributes of the user from the user's user profile from the social media system 140 with web scraping or by other means of an application programming interfaces (API) to the social media system 140. This may require the user to provide authorisation for the user profile extractor to download the user's user profiles, e.g. by logging in to the social media system 140 in question and setting a certain setting.

Figures 6a-d show examples of extracting user's attributes in user profiles from social media systems 104 by means of screen scraping and web scraping techniques. Figure 6a shows a user interface 600 as may be provided by the analytics server 1 10 for output via the client running on the user terminal 122. The user interface 600 enables the user 120 to specify the type of insurance cover required 620 and to fill in his/her contact details 630. There is also provided a clickable button 610 for the user to authorise the user profile extractor at the analytics server 1 10 to contact one or more social media systems 140 and therefrom to extract his/her user profile, or selected information therefrom. Upon clicking the relevant button 610 and selecting a social media system 140, the user is directed to an authorisation interface 602 for inputting login credentials 650, as shown in Figure 6b. Said authorisation interface 602 also discloses 640 the attributes of the user in the user profile that are provided to the user profile extractor. In the given example, the user has selected Linkedin as their preferred social media system where the attributes of the user in the user profile available for download include name, photo, headline, current positions and email address.

In embodiments, the user will be taken to a profile summary page 660 which shows various attributes of the user in the user profile on-screen. In some embodiments, the profile summary page 660 as shown in Figure 6c is available for public view and thus does not necessarily require authorisation from the user. In this case, the login credentials 650 inputted are for identifying the correct user profile on the social media system. In some embodiments, the profile summary page 660 is a private profile page and the attributes of the user in the user profile shown thereat are not open to public, as it may contains private or security sensitive aspects of the user profile, e.g. date of birth and national insurance/social security numbers.

At the public profile page 660 in Figure 6c, the user profile extractor identifies the required attributes of the user in the user profile, e.g. based on text analysis or via a suitable API. For example, the user profile extract recognises keywords 662 "Senior programmer" and "Information Technology and Services" from the profile page 660 and therefrom identifies one or more aspects of the user profile, e.g. the user's occupation, industrial section and seniority in the company. Similarly the workplace location 664 of user may be readily extracted from the profile page 660. These attributes of the user in the user profile affect insurance premium estimation when they are analysed and outputted as analytics outputs, to subsequently be applied in the estimation algorithms. For example, a mid-age senior programmer working in London posts significantly lower risk than an oil rig operator of an advanced age working in the North Sea. Other attributes of the user in the user profile extractable by screen scraping or other API means the profile page 660 include the size of social circle 666, work experience 668, technical background and level of education 670. In some embodiments, the user profile extractor may also be able to search the meanings of complex keywords extracted from the profile summary page 660 at a search engine or database and thereby output one or more attributes of the user in the user profile. For example, upon identifying the keywords "Unity 3D", "Unreal Engine 4", "Gameplay" and "AI", the web profile extractor consults a search engine such as Google.com for the meaning of such keywords. It then analyses the returned search results and identifies the keywords are commonly associated with the "Information Technology and Services" industry, and thereby registers an analytics output at the user profile database 1 14. In embodiments where a web scraping technique or other API is used, the extracted aspects of the user profile need not necessarily be shown on screen. Once the user has successfully logged into the selected social media system, authorisation is given to the user profile extractor to download the available attributes of the user in the user profile directly from the social media systems 140, e.g. attributes of the user described above in the profile summary page 660, and/or user's attributes in the user profile that are not accessible by the public. Advantageously, this does not require personal or security sensitive attributes of the user in the user profile to be displayed on the user screen and therefore significantly enhances the security of system, because a malicious party may no longer physically view sensitive information or attributes displayed on screen.

Once the one or more attributes of the user in the user profile are extracted by either the screen scraping, web scraping or other API based technique, the user is directed back to the interface 604 as shown in Figure 6d. The one or more attributes of the user in the user profile(s) are automatically filled into the relevant fields, e.g. user details such the name, date of birth and email address of the user are shown filled in the interface 604. The processing apparatus 1 12 is configured to, based on analysing the available attributes of the user, output analytics outputs to fill in the relevant fields 624 at the interface 604. For example, using the date of birth, gender and/or occupation, or such like, downloaded from the social media system 140, the processing apparatus 1 12 may determine the user's life expectancy. Furthermore, based on analysing the occupation and position of the user, the processing apparatus 1 12 may estimate annual income of said user and anticipates the amount and level of insurance cover required.

The one or more attributes of the user in the user profile may be basic statistical data such as age, gender, occupation, place of residence, income and/or education, e.g. these basic statistical data are readily downloadable without further analysis. However, such readily downloadable attributes of the user in the user profile may not be sufficient for getting an insurance estimate. Therefore one or more analytics outputs may be generated by further processing of telemetry data or photos. For example, the user may have logged all the places he/she had visited in the past year, and based on the geographical location of each of the visited places, the user profile extractor may extract the user's travel itinerary and therefrom output analytics outputs such as the number of flights the user has taken in a year, any high risk areas the user has visited, e.g. a conflict zone, or any high risk activities the user has participated in, e.g. frequent visits to ski resorts indicates that the user is a keen skier. In some embodiments, the one or more attributes of the user in the user profile may comprises social aspects of the user. For example the processing apparatus 1 12 may extract the number of the connections the user has so as to determine the size of his social circle, or the processing apparatus 1 12 may analyse the interaction the user has with his/her connections, e.g. the frequency of "likes" given and/or comments the user has made or received. In some embodiments, the processing apparatus 1 12 may analyse one or more attributes of the user in the user profile (e.g. age, education and occupation) of at least some of the user's connections, in order to determine the circle of friends the user keeps. This enables the profile extractor to output one or more analytics outputs that are missing from the social media system 140. For example, by analysing the one or more attributes of the user in the user profiles of the user ^' s connections, the profile extractor identifies skiing is a common hobby shared by the user's connections, and thereby estimates that the user may also be a skier even though the user profile does not specify skiing as a hobby.

Any given one of the estimation algorithms maybe configured to use any one or more of the above factors as inputs based upon which the algorithm generates its respective estimate at the analytics server 1 10. In some embodiments, the user may have photos uploaded and stored at the social media system 140, e.g. pictures stored in various albums accessible by "everyone" or trusted users. One or more of these stored photos maybe downloaded by an attributes extractor and stored at the user profile database 1 14, so that they can be analysed by any suitable image recognition algorithm in order to recognise and extract the one or more analytics outputs. Using the example given above, and as illustrated in Figure 7, the user may have holiday photos 160 stored at the social media system 140 that are accessible by the attribute extractor. The processing apparatus 1 12 may then carry out image recognition process on the stored photos to identify different features (e.g. activity, clothing, background) therein. For example, the image analysis recognises an activity "skiing" from the user's skiing posture, as well as recognising the user is wearing skiing gear 162 by analysing said user's clothing. Moreover, image analysis recognises a white backdrop 164 as a snowy mountain. Thus by combining all these recognised states of features the process apparatus 1 12 deduce one or more analytics outputs, e.g. skiing as hobby and/or activity the user has participated in. Other examples that can be recognised by features in a user's photos from the social media systems 140 may include the user's lifestyle. For example, the processing apparatus 1 12 may recognise the presence of cigarettes or wine glass in the user's photos and/or their frequency of occurrence, and therefrom estimates the likelihood said user is a heavy smoker or an alcoholic as an analytics output.

In some embodiments, the social aspects of the user 120 may also be investigated by analysis of the user's photos stored at the social media systems 140. For example the processing apparatus 1 12 may analyse how the user 120 interacts with his/her connections, e.g. by recognising the how frequent the user 120 has "tagged" or been "tagged" by his/her connections, as well as what activities they undertook during said interactions.

In an alternative or additional embodiment, the user 120 may use an optical camera on-board the user terminal 122 to capture an image of himself/herself for use with the image recognition technique mentioned above. Said image may be a facial image or an image showing the user's bodily features so to enable the image recognition technique to recognise and evaluate the state of user's visual features and therefrom determine one or more aspects of the user, e.g. age and/or health condition. Note that said facial image or the image showing the user's bodily features as captured by the on-board camera may be processed in the same way as user's stored photos accessible at the social media website.

In some embodiments, the user 120 may capture a facial image of himself/herself, such that the user's facial features may be analysed to evaluate the state of said facial features, and therefrom determine one or more analytics output. For example, the facial image of the user can be broken down into several facial features, e.g. wrinkles, hair quantity/colour and moustache. E.g., as detected in the facial image, the user has exhibited a substantial amount of wrinkles at a specific position (e.g. forehead), as well as a head of grey/white hair. Based on these two facial features the processing apparatus 1 12 may be able to estimate the user's age, or at least categories the user as an aged user.

Other aspects of user profile that are determinable from analysing user facial features or bodily features may include gender, as men have a lower life expectancy than women. For example, the processing apparatus detects a moustache in the user's facial image and thereby confirms that the user is a male. In some embodiments, the user's facial feature may be used for assessing the user's lifestyle. For example the processing apparatus 1 12 may recognise a yellowish tint in the user's teeth and therefrom categorising the user as a heavy smoker as an analytics output. Or in some other cases where the processing apparatus 1 12 recognises a number of scars on the user's facial image, and therefrom determines that the user may have actively involved in extreme sports, or some other dangerous activities.

In some embodiments, the user's health condition, as one or more analytics outputs, may be recognised from his/her facial and/or bodily image(s). For example, the user's body mass index (BMI) can be roughly estimated from the height and the user's on-screen presence. In some embodiments, the one or more analytics outputs comprises sightedness, hearing ability, and/or degree of physical impairments and/or other health issues, which may be recognised from the facial images and/or bodily images. For example, the processing apparatus 1 12 may detect the outline of a pair of glasses or hearing aids in the facial image and/or a bodily image of the user. Upon evaluating the state of such features the processing apparatus 1 12 summarised in an analytics output of the user profile, that said user requires sight correction or hearing aids and therefore may need to spend more money on these items.

In some embodiments, the analytics server 1 10 may cross-reference one or more of the analytics outputs so as to verify the authenticity of the data provided by the social media systems 104. For example if the age of the user is known from other sources, e.g. social media system, it can be cross referenced with the evaluated states of one or more facial features. For example if the user has a reported age of 25 and the processing apparatus 1 12 determines, from the wrinkles and grey hair, the user 120 is actually at an advanced age, the processing apparatus 1 12 may deduce that the user must either i) have provided fraudulent information or attributes in his/her user profile at the social media systems 140 or ii) be of ill health or be a heavy smoker and/or alcoholic. In some embodiments, the processing apparatus 1 12 may then require the user to provider further information, e.g. proof of age or further information on health conditions. In some embodiments, the photo as supplied by the user (e.g. captured by an onboard camera) may be compared with one of the user's photos stored at the social media website, in order to verify the genuineness of the supplied picture. Using the above example, the user in his advanced age may provide a fraudulent picture featuring a younger looking man, in an attempt to lower his insurance estimate. Upon comparing the supplied picture with a randomly selected photo stored on the social media website, the processing apparatus 1 12 determines there is a discrepancy between the user's appearance in the supplied photo and the stored image, thereby deduces that the user must have provided fraudulent information. This method prevents user from submitting a fraudulent picture, or using someone else's social media account, in order to achieve a lower insurance estimate. In some embodiments, upon detecting fraudulent user information, e.g. discrepancy between the reported age and/or reported gender of a user and the appearance of the user, the analytics server 1 10 notifies the social media systems of said fraudulent user information or attributes via the API so to reduce the number of bogus user accounts used by predators. In some embodiments, a video stream that is either downloadable at the social media systems 104 or captured by the on-board camera at the user's terminal 122, may be used in place of the photo for recognising user's facial features and/or bodily features. For example, the video stream may be broken down in to a plurality of frames (or images) each featuring at least one of facial features and/or bodily features of the user, such that the image recognition technique as mentioned above can be applied to at least one of the plurality of frames. In some embodiment, a video recognition technique may be used to recognise the changes in states in user's features over the plurality of frames in the video, so to determine one or more analytics outputs of the user profile. As an example, the video recognition technique may recognise the user has restricted movement in a video feed and therefrom deduces that the user may have a certain health issue. Or in the earlier case, the video recognition technique may analyse a footage capturing the user's skiing actions and recognises the user is travelling at a reckless speed, thereby categorising the user as a high risk skier as an analytics outputs.

Once the one or analytics outputs of the user profiles are determined they are stored in the user profile database 1 14. The processing apparatus 112 may then estimate a plurality of insurance premiums based on the determined analytics outputs and the estimation algorithms provided by the respective insurance providers. In some cases more than one estimation algorithms may be provided by the same insurance provider, each for calculating estimates for a different product, e.g. different estimation algorithms may be provided for each of third party only insurance and fully comprehensive insurance. The plurality of insurance estimates are stored at the estimates database 118 which may be retrievable by said user at a later date.

Finally, the plurality of insurance estimates are provided in a list to the user. Preferably, the plurality of insurance estimates may be ranked according to their cost or any other factors, such as relevance, ratings of the insurance provider or insurance product, or promotional discounts. The ranked list of estimates is then provided, on screen and/or by other means such as email, to the user along with identities of the respective insurance providers. In some cases, hyperlinks are provided to direct the user to the service providers' websites for purchasing the insurance products.

It will be appreciated that the above embodiments have been described by way of example only. Other variants and/or applications of the disclosed techniques may become apparent to a skilled person once given the disclosure herein. The scope of the present disclosure is not limited by the above-described embodiments but only by the accompanying claims.