IE20050764U1 - Mobile device internet access - Google Patents

Abstract

ABSTRACT A mobile device browser accesses (2) a proxy server and the user selects preferred categories for web pages. This is done using a graphical interface generated by a feedback mechanism of the proxy server. A profiling mechanism of the proxy server incorporates an artificial intelligence engine and it maintains a table of preferred categories in a customer profile. The mobile device accesses (3) a recommended site. The mobile operator server downloads (4) instructions to the mobile device to cause it to display a toolbar. The user selects one icon on the toolbar to indicate whether the site is approved or disapproved. If the approval icon is clicked, the server automatically downloads (5) a page for posting comment. The page or form includes a list of options to be ticked by the user to indicate the preferred subject matter areas such as science, ring tones. entertainment, sport, technology, and news. Also, the automatically transmits (6) notifications to other users by using a current database of approved sites. If disapproved, the above steps are repeated for a next site. The feedback mechanism of the server is configured to enable transformation of mobile content by addition of the toolbar. The profiling mechanism of the server performs real time monitoring of content access by mobile devices, and determines if it scores highly enough to be successful.

Description

The invention relates to access by mobile devices to the intemet.

Prior Art Discussion In recent years there have been significant developments in mobile device access to the intemet. However a major restriction is, by necessity, the small size of the keypad and the display screen. This tends to limit the extent to which users access web pages using the mobile device browser. Instead, they tend to use only a limited number of specific sites.

The invention is directed towards achieving more convenient access to relevant web pages by browsers executing on mobile devices.

SUMMARY OF THE INVENTION According to the invention there is provided a mobile network proxy server comprising:- an interface to receive content access requests from a wide area network server and originating in mobile devices; a feedback mechanism for generating a graphical interface suitable for display on a particular mobile device, for transmitting the interface to the mobile device, and for receiving in real time user feedback within the structure of the graphical interface; and [E 050764 a profiling mechanism for dynamically maintaining a user content preference profile according to the user feedback, and for automatically determining starting content upon receipt of first accesses by mobile devices.

In one embodiment, the feedback mechanism automatically generates a toolbar graphical interface and it recognises user selections on the mobile device of buttons on the toolbar as particular quantified rating.

In one embodiment. the feedback mechanism extracts mark-up text and transmits said text to a mobile device, and the mobile device uses the mark-up text to generate the toolbar graphical interface.

In one embodiment, the feedback mechanism automatically determines mobile device type and transmits the mark-up text accordingly.

In one embodiment, the toolbar has a discrete icon for indicating approval of content such as a site.

In one embodiment, the server has a discrete icon for indicating disapproval of content such as a site.

In one embodiment, the feedback mechanism automatically downloads a graphical interface for user input of more detailed feedback, upon receipt of a positive rating.

In another embodiment, the feedback mechanism modifies mobile content browser- accessible pages to display the user prompts.

In one embodiment, the feedback mechanism parses the responses in real time to modify the user prompts.

In one embodiment, the feedback mechanism transmits the graphical interface to the mobile device with instructions to display the graphical interface irrespective of which content is accessed.

IE 050764 In one embodiment, the profiling mechanism manages a list of approved content sites, as indicated in feedback from multiple users.

In one embodiment, the profiling mechanism dynamically monitors data traffic and downloads an indicator for display on the toolbar.

In one embodiment, the profiling mechanism manages a database of mobile device browser capabilities and dynamically manages downloads to the mobile device accordingly.

In a further embodiment, the profiling mechanism classifies content sites according to category and weight.

In one embodiment, the profiling mechanism maintains a look-up table of weights and uses random values in the table for an initial seeding process to identify a starting content site.

In one embodiment, the profiling mechanism normalizes the number of user selections in a site to provide a weighting component.

In one embodiment. the profiling mechanism normalizes a time value for time spent by a user visiting a content site to provide a weighting component.

In one embodiment, the profiling mechanism comprises a recommendation artificial intelligence engine to automatically generate a content recommendation.

In one embodiment, the engine uses feedback from a plurality of users to generate a content recommendation for a particular user.

In one embodiment, the engine comprises a knowledge base and an inference engine which automatically interprets knowledge base rules. lE05oz64 DETAILED DESCRIPTION OF THE INVENTION Brief Description ofthe Drawings The invention will be more clearly understood from the following description of some embodiments thereof, given by way of example only with reference to the accompanying drawings in which:- Fig. l is a flow diagram for mobile device operations according to the invention; Fig. 2 is a representation of a toolbar; Fig. 3 is a flow diagram for server operations according to the invention, and F i g. 4 is a more detailed flow diagram of the server operations.

Description of the Embodiments Referring to Fig. 1 in a method 1, a mobile device browser accesses a proxy server in a step 2 and the user selects preferred categories of web pages. This is done using a graphical interface generated by a feedback mechanism of the proxy server. A profiling mechanism of the proxy server maintains a table of preferred categories in a customer profile. In a step 3 the mobile device browses to a recommended site. In step 4, the mobile operator server downloads instructions to the mobile device to cause it to display a toolbar such as shown in Fig. 2. The user selects one icon on the toolbar to indicate whether the site is approved or disapproved. If the approval icon is clicked, the server automatically downloads a page for posting comments, in step 5.

The page or fomi includes a list of options to be ticked by the user to indicate the preferred subject matter areas such as science, ring tones, entertainment, sport, technology, and news. Also, the server automatically transmits notifications (step 6) to other users by using a current database of approved sites. If disapproved, steps 3 onwards are repeated for a next site. lEo5o754 The server operations are illustrated in Fig. 3 as a method 20. In step 21, the feedback mechanism of the server is configured to enable transformation of mobile content by addition of the toolbar of Fig. 2. In step 22 recommended starting sites are set, and content is launched in step 23. The profiling mechanism of the server performs real time monitoring of content access by mobile devices, and in step 25 determines if it scores highly enough to be successful. A next site or other item of content is chosen in step 26.

The operations are described in more detail as a method 30 in Fig. 4 having steps 31 to 44. In step 3] the mobile device uploads a starter URL and in step 32 the WAP gateway directs the device to the proxy server, which in step 33 attempts to connect to a recommended site. If such a site is not found, as indicated by steps 34 and 44, the proxy server profiling mechanism processes database records to identify alternative sites and returns to step 32.

When a site is found, in step 35 the proxy server feedback mechanism identifies the mobile device from the header information in step 36, and in step 38, it inserts mark- up text to generate the toolbar at the correct location within mark-up text. It also (step 39) identifies and modifies all links and tags, including image and meta tags such that they are compatible with proxy server and toolbar insertion. The mobile device then generates and renders the mark-up text.

In step 40 the mobile device renders the mark-up text, and in step 41 the user selects a button on the toolbar. The proxy server, in step 42, updates the preferences to a persistent store 43. As indicated by step 44 the proxy server may then identify a further site.

In more detail, referring again to step 22 of Fig. 3, the proxy server profiling mechanism transmits content that users they find personally compelling by building up a profile of that user's preferences over time. When a user first accesses the Web server the proxy server cannot determine what that user’s preferences are because there is no data concerning this user. The profiling mechanism performs a seeding [E 050764 step by taking the most popular and most liked content from other users and serves it to the new user. This is done by querying the profile database for live sites. The live sites are ordered by category and weight. The weight is calculated by taking a seed value which is multiplied by a value chosen by the mobile operator in the range of 0.1 to 0.8. This combined value is multiplied by the category weight value which is determined from a look-up table of values which are created randomly initially and then based on usage refined to represent weights based on continuing use. This ensures that the initial random weightings seek accurate values very quickly. In addition the total number of clicks made by users within a site is normalized to a value between 0.1 and 1 and this value is multiplied by the width so that sites that have had more clicks by users get a higher weight. Similarly, the time spent on sites is normalized to values between 0.1 and 1 and this value is multiplied by the weight so that sites that users spend more time on get a higher weight. Finally a random value is generated between 0.1 and l and this value is multiplied by the weight so that the sites with the higher weights tend to get displayed first but at any one time a low weight site could be displayed within the first few sites so that users see the best sites first and sometimes see a lower weight site also. This ensures that sites with a lower weight are seen by users and get a chance to have a higher weight associated with them.

Also, for step 44, the proxy server executes a process that weights the collaborative recommendation of other users and prompts these recommendations to the appropriate other users. The process produces a metric that measures the most suitable content for each individual user based on recommendations of similar users.

The collaborative recommendation is handled by an artificial intelligence engine.

This is a dedicated expert system limited to solving problems and giving advice within the specialized knowledge domain of the proxy server and registered mobile devices.

The artificial intelligence engine analyses user patterns and calculates in real-time the best content to transmit to a particular user. The basic components of the expert system are a knowledge base (or “KB”) and an inference engine. The information in the KB is obtained by eliciting decision—making rules from the base of all previous IE 050764 decisions made by end-users. The information obtained is organized into a collection of rules, (if—then structured, with an attached probability weighting for each rule).

These rules (production rules) are fired depending on the inputs supplied by a query sent to the server (including current site, user's last ten sites, user category selections, user likes, user dislikes). The inference engine enables the expert system to draw deductions from the rules in the KB. The deductions each have an attached degree of truth co-efficient (i.e. fuzzy values). The deductions are combined by a voting system to return the most appropriate site for the end user.

The proxy server may send to the mobile device user in a pseudo random fashion content sites that they are likely to like based on previously recorded usage patterns of other users of the service in the portal. A collaborative rating approach ensures this content is best suited to this individual user.

The feedback mechanism of the proxy server can convert graphics that would not normally be displayed in the browser into a format that is supported. This is based on a database of 4,000 handsets and their individual graphic rendering capability. In a similar manner, the markup language of the visited site can be dynamically changed by the server into a mark-up language supported by the browser.

Through dynamic transformation of the pages rendered on the mobile device, the user has the option to: - read the mobile intemet or content site normally and continue browsing, v rate it positively, — rate it negatively, - add a comment about the site, - view cost of the browsing session (calculated and displayed in real time), or — go to the next random site.

At any time a user can modify his list of interests to include or omit topics. For example, an alphabetical list may include topics ranging from Astronomy to Zoology.

[E 050764 An important result is that the mobile device user is always within control of the system operator and is always just one browser click from the home portal as operated by the mobile network operator. In effect, the mobile browser is free to roam outside of the mobile operator portal but never leaves the portal as the persistent toolbar is always available.

It will be appreciated that the proxy server executes a process that weights the collaborative recommendation of other users and prompts these recommendations to the appropriate other users. This process produces a metric that measures the most suitable content for each individual user based on recommendations of similar users.

The proxy server modifies mobile content browser accessible pages from arbitrary sites on-the-fly so that the relevant and available options are displayed as extra clicks on the top or bottom of the page. The page is parsed in real time and extra clicks are added to the top or bottom (or both). This requires the raw mark up language to be modified and extra code inserted to add via a toolbar.

The proxy server builds up an initial database of usage patterns for initial launch in a network. The system relies on people using it to be successful and this process needs to be bootstrapped. Once the usage patterns have been bootstrapped a process is executed to detect relevance and usage patterns that will direct other users to content that they are genuinely interested in.

The proxy server filters out sites that are not accessible by other users. For example, if a user recommends a content page of a mobile intemet site that requires a subscription these should not be pushed to other users. This should happen naturally if the above process is effective.

When a user clicks a button on the toolbar that indicates positive approval of a site that has already been added to the system, various calculations are made based ion this usage pattern. When a user clicks this approval button on a site that is not in the system it is added to a pending list. If another user clicks the approval button on the same site it remains on the pending list but the priority is increased. A System operator visits the list of pending over Web interface and performs two tasks: lEo5o764 - Categorisation. Add the site to one or more content categories based on the text and images in the site.

- Approval. Approve or reject the site. Operators may wish to exclude certain sites altogether.

The system also makes it possible to completely automate the approval/ categorisation process. All sites added are in an approved state by default (can still reject later) and categories are guessed from the content of the site (categories can still be changed later). The Auto Categorisation relies on each category having a list of search terms associated with them and an intelligent matching algorithm.

The operator of the system has full control over the database so that certain sites can be blocked and others pushed.

It will be appreciated that the invention enables the operator to visualize and map the consumption of mobile content made available through the toolbar. A user can recommend content to others: when users click on the toolbar to approve content they can post messages to other users via SMS/e—mail/live chat or on their own web log portal. Direct communication with other users is possible in this way. A user can view the sites recommended by other users and history of sites visited using the system. A user can bookmark sites approved of on a remote server for retrieval via the mobile device.

The invention is not limited to the embodiments described but may be varied in construction and detail.

Claims (1)

1. Claims [E 050764 A mobile network proxy server comprising:- an interface to receive content access requests from a wide area network server and originating in mobile devices; a feedback mechanism for generating a graphical interface suitable for display on a particular mobile device, for transmitting the interface to the mobile device, and for receiving in real time user feedback within the structure of the graphical interface; and a profiling mechanism for dynamically maintaining a user content preference profile according to the user feedback, and for automatically determining starting content upon receipt of first accesses by mobile devices; wherein the feedback mechanism automatically generates a toolbar graphical interface and it recognises user selections on the mobile device of buttons on the toolbar as particular quantified rating; wherein the feedback mechanism extracts mark-up text and transmits said text to a mobile device, and the mobile device uses the mark-up text to generate the toolbar graphical interface; and wherein the feedback mechanism automatically determines mobile device type and transmits the mark-up text accordingly. A mobile network proxy server as claimed in claim 1, wherein the toolbar has a discrete icon for indicating approval of content such as a site, wherein the toolbar has a discrete icon for indicating disapproval of content such as a site; and wherein the feedback mechanism automatically downloads a graphical interface for user input of more detailed feedback, upon receipt of a positive rating. A mobile network server as claimed in claims 1 or 2, wherein the feedback mechanism modifies mobile content browser-accessible pages to display the user prompts; wherein the feedback mechanism parses the responses in real time to modify the user prompts; wherein the feedback mechanism transmits the graphical interface to the mobile device with instructions to display the graphical interface irrespective of which content is accessed; wherein the profiling mechanism manages a list of approved content sites, as indicated in feedback from multiple users; and wherein the profiling mechanism dynamically monitors data traffic and downloads an indicator for display on the toolbar. A mobile network proxy server as claimed in any preceding claim, wherein the profiling mechanism manages a database of mobile device browser capabilities and dynamically manages downloads to the mobile device accordingly; wherein the profiling mechanism classifies content sites according to category and weight; wherein the profiling mechanism maintains a look-up table of weights and uses random values in the table for an initial seeding process to identify a starting content site; wherein the profiling mechanism normalizes the number of user selections in a site to provide a weighting component; wherein the profiling mechanism normalizes a time value for time spent by a user visiting a content site to provide a weighting component; wherein the profiling mechanism comprises a recommendation artificial intelligence engine to automatically generate a content recommendation. A mobile network proxy server as claimed in claim 4, wherein the engine uses feedback from a plurality of users to generate a content recommendation for a particular user; wherein the engine comprises a knowledge base and an inference engine which automatically interprets knowledge base rules.