WO2014022893A1

WO2014022893A1 - Interface for visually navigating concepts

Info

Publication number: WO2014022893A1
Application number: PCT/AU2013/000880
Authority: WO
Inventors: Adam Luke TOMAS
Original assignee: Tomas Technology Pty Ltd
Priority date: 2012-08-09
Filing date: 2013-08-09
Publication date: 2014-02-13

Abstract

A method for visually navigating concepts on a computing device, including: displaying selectable visual objects representing concepts in central and peripheral regions of an interface rendered on a display of the computing device; receiving input representing one or more selections of one or more visual objects in the peripheral regions representing one or more selected concepts; updating the central region of the interface to display the one or more selected visual objects; and updating the peripheral regions of the interface to display new visual objects representing new concepts related to the one or more selected concepts.

Description

INTERFACE FOR VISUALLY NAVIGATING CONCEPTS

Field

[0001] The present invention relates to an interface for visually navigating concepts on computing devices, for example, smartphones, tablets, laptop computers, and desktop computers.

Background

[0002] Concepts are conventionally navigated on displays of computing devices using search engines that use textual search queries to return search results as long lists of links each leading to a web page or web document. Navigating concepts using search engines involves the time-consuming and cumbersome process of examining long lists of search results, retrieving the search results, repeatedly reformulating and resubmitting search queries, and trying different ways to express concepts of interest. Conventional search engines therefore place the burden of defining, seeking, extracting and organising concepts on the user.

[0003] A need therefore exists for solutions for navigating concepts on computing devices that address or alleviate at least some of the problems described above.

Summary

[0004] According to the present invention, there is provided a method for visually navigating concepts on a computing device, including:

displaying selectable visual objects representing concepts in central and peripheral regions of an interface rendered on a display of the computing device; receiving input representing one or more selections of one or more visual objects in the peripheral regions representing one or more selected concepts;

updating the central region of the interface to display the one or more selected visual objects; and

updating the peripheral regions of the interface to display new visual objects representing new concepts related to the one or more selected concepts. [0005] The method may further include receiving input representing one or more deselections of one or more visual objects in the central or peripheral regions representing one or more deselected concepts, and updating the peripheral regions to display new visual objects representing new concepts related to the one or more selected concepts.

[0006] The method may further include receiving input representing two or more selections of two or more visual objects in the peripheral regions representing two or more selected concepts, updating the central region to display the two or more selected visual objects, and updating the peripheral regions to display new visual objects representing new coricepts having a compound relationship to the two or more selected concepts.

[0007] The two or more selected concepts may be summarised in the central region as a tree structure or list representing a concept hierarchy.

[0008] The new visual objects may be displayed automatically based at least in part on relevance ratings between the new concepts and the selected concepts. Additionally, the new visual objects may be displayed automatically based at least in part on randomising the relevance ratings between the new concepts and the selected concepts.

[0009] The central and peripheral regions of the interface may be arranged in a two- or three-dimensional visual representation selected from a group comprising a matrix a cluster, a grid, and an array.

[0010] The visual objects may be interactive media items such as thumbnails or icons that visually represent content or physical items.

[0011] The present invention also provides a graphical user interface (GUI) for visually navigating concepts on a computing device, including:

displaying selectable visual objects representing concepts in central and peripheral regions of the GUI rendered on a display of the computing device; receiving input representing one or more selections of one or more visual objects in the peripheral regions representing one or more selected concepts;

updating the peripheral regions of the GUI to display new visual objects representing new concepts related to the one or more selected concepts.

[0012] The present invention further provides at least one computer-readable medium having stored thereon instructions for visually navigating concepts on a computing device, the instructions being executable to cause a computer processor to:

display selectable visual objects representing concepts in central and peripheral regions of an interface rendered on a display of the computing device; receive input representing one or more selections of one or more visual objects in the peripheral regions representing one or more selected concepts;

update the central region of the interface to display the one or more selected visual objects; and

update the peripheral regions of the interface to display new visual objects representing new concepts related to the one or more selected concepts.

Brief Description of Drawings

[0013] Embodiments of the invention will now be described by way of example only with reference to the accompanying drawing, in which:

Figure 1 is a flowchart illustrating a method for visually navigating concepts according to an embodiment of the invention;

Figures 2 to 5 illustrate example interfaces for visually navigating movies;

Figures 6 and 9 illustrate example data structures of the method;

Figures 7 and 8 illustrate example interfaces for visually navigating music;

Figures 10 to 17 illustrate example interfaces for visually navigating travel planning;

Figures 18 to 22 illustrate example interfaces for visually navigating online grocery shopping; and Figure 23 illustrates an example interface for visually navigating clothes shopping.

Detailed Description

[0014] Figure 1 is a flowchart illustrating a process 10 for visually navigating concepts on a computing device according to an example embodiment of the present invention. The computing device is, for example, a smartphone, a tablet, a laptop computer, or a desktop computer. Concepts are abstractions of real world objects, situations, events or patterns. At step 12 an initial cluster of user selectable and interactive visual objects representing concepts is displayed in central and peripheral regions of a graphical user interface (GUI or "interface"). The cluster of visual objects visually represents a "concept cluster". The visual objects are interactive media items such as thumbnails or icons that visually represent content or physical items. Descriptive text or metadata related to the visual objects is optionally displayed in the user interface in or adjacent the visual objects. The central region is optionally defined in the interface by a frame or window. .

[0015] The visual objects in the initial concept cluster are identified, populated and displayed in the interface based on one or more of a user search query, a suggestion based on a user browsing history, a user search history, a user profile, or a redirection from a browser, application or page. Visual objects are selected and deselected by mouse or touchscreen input, such as pointing to, clicking on or touching visual objects, or dragging and dropping visual objects from the peripheral regions to the central region and vice versa. The central and peripheral regions of the interface are, for example, arranged in a two- or three-dimensional visual representation selected from a group comprising a matrix, a cluster, a grid, and an array.

[0016] At step 14 user input is received by the interface representing one or more selections of one or more visual objects in the peripheral regions representing one or more selected concepts. The one or more visual objects are selected by, for example, pointing to, clicking on or touching, or by dragging from the peripheral regions and dropping in the central region. [0017] Responsive to the one or more user selections at step 14, the central region of the interface is updated dynamically at step 16 to display the one or more selected visual objects, and the peripheral regions of the interface are updated dynamically at step 18 to display new visual objects representing new concepts related to the one or more selected concepts. If two or more visual objects in the peripheral regions are selected at step 14 representing two or more selected concepts, the central region is updated dynamically at step 16 to display the two or more selected visual objects, and the peripheral regions of the interface are updated dynamically at step 18 to display new visual objects representing new concepts having a compound relationship to the two or more selected concepts. The two or more selected visual objects (or their corresponding descriptive text) are optionally summarised in the central region in a concept hierarchy, for example, a tree structure or list that represents an order of selection of the two or more visual objects representing the two or more selected concepts.

[0018] Alternatively at step 14 user input is received by the interface representing one or more de-selections of one or more visual objects in the central or peripheral regions representing one or more deselected concepts. The de-selection of visual objects represent user rejection of the visual objects and corresponding concepts displayed in the central or peripheral regions. For example, clicking on or touching the central region optionally deselects the initial cluster of visual objects displayed in the central and peripheral regions at step 2. One or more visual objects displayed in the central region are deselected by, for example, dragging outside the central region and dropping in peripheral regions. In response to de-selection of visual objects, the peripheral regions of the interface are updated dynamically at step 8 to display new visual objects representing new concepts related to the one or more selected concepts displayed in the central region. Additionally or alternatively, clicking in the central region rejects all of the suggested new concepts in the peripheral regions, and the peripheral regions are subsequently updated and re- populated with visual objects representing fresh new concepts related to the one or more selected concepts displayed in the central region.

[0019] As described below the new visual objects are identified and dynamically displayed automatically in the interface based at least in part on relevance (or similarity) ratings between the selected concepts and the new concepts. The relevance ratings are based at least in part on mapping contextual or semantic relationships between the selected concepts and the new concepts. The relevance ratings are calculated using contextual or semantic mapping techniques using conventional logic, statistical methods or mathematical algorithms. To enable a user to discover new concepts, the relevance ratings may be at least partially randomised to provide selective degrees of serendipitous discovery of new concepts that are related in varying random degrees of relevance to the original or selected concepts. The relevance and randomisation functionalities are implemented respectively by conventional relevance and randomisation techniques, for example, neural networks, probability trees, Bayesian probability techniques, etc.

[0020] Steps 14 to 18 are repeated, and the process 10 is performed iterativeiy, to allow a user to dynamically navigate concepts by visually interacting with visual objects in the central and peripheral regions of the interface.

[0021] The process 10 is implemented by computer code stored on a computer- readable medium that causes a device to perform steps 12 to 18. The device is, for example, a personal computer, a laptop, a tablet, a smart phone or a mobile phone.

[0022] The invention will now be described in more detail by reference to the following examples, which are illustrative only and do not limit the scope of the invention.

Example 1 - Movies

[0023] Figure 2 illustrates an example graphical user interface 20 provided by the process 10 for conceptually navigating and exploring a database of movies. The interface 20 is provided as a 3x3 two-dimensional matrix of movie thumbnails. Other equivalent two- or three-dimensional clusters of visual objects may also be used, such as circular or triangular arrays, or cubes. A thumbnail representing The Godfather movie (illustrated as a frame with descriptive text for clarity) is displayed in a central region 22 of the interface 20 as an initial concept of interest based on a web search query, a suggestion based on web history or user profile, or a redirection from a web browser or web page. An initial cluster of thumbnails of related movies is displayed in peripheral regions 24 on the interface 20 around the central region 22. Each thumbnail is interactive by the interface 20 to allow a user to interact with media items or content associated with the corresponding movie. The interaction includes, for example, zooming in on the thumbnail of the movie, reading an online review of the movie, going to a website for the movie, purchasing the movie, downloading and watching the movie, saving the movie for subsequent interaction, etc.

[0024] The interface 20 provides a number of options for dynamically navigating to subsequent clusters of related movies. In a first option, a user explores movies related to one of the movies in the peripheral regions 24 by pointing to and clicking on one of the peripheral movie thumbnails. The interface 20 then moves and dynamically displays the selected peripheral movie thumbnail in the central region 22, and dynamically displays new thumbnails of related movies in the peripheral regions 24. Figure 3 illustrates the interface 20 after user selection of The Deer Hunter movie in a peripheral region 24 of Figure 2. The central region 22 of the interface is dynamically updated to display the thumbnail of The Deer Hunter movie, and the peripheral regions 24 are dynamically updated to display thumbnails of movies related to The Deer Hunter.

[0025] In a second option, the user rejects or discards ail the related movies Initially displayed in the peripheral regions 24 of Figure 2 by pointing to and clicking in the central region 22. As illustrated in Figure 4, the peripheral regions 24 of the interface 20 are dynamically updated to display a new cluster of thumbnails of movies related to The Godfather movie displayed in the central region 22.

[0026] In a third option, the user explores movies having a compound relationship to the movie represented by the thumbnail displayed in the central region 22 and one or more of the related movies represented by the thumbnails displayed in the peripheral regions 24. Compound relationships are explored via the interface 20 by clicking on one of the related movie thumbnails in the peripheral regions 24 and dragging it into the central region 22. As illustrated in Figure 5, the central region 22 of the interface 20 is updated dynamically to display thumbnails of both the original movie The Godfather and the selected additional movie The Deer Hunter dragged inwardly from the peripheral regions 24. The peripheral regions 24 are updated dynamically to display thumbnails of movies that are jointly related to the two movies in the central region 22.

[0027] The user selectively repeats the above navigation options (as well as discarding a concept from a compound concept in the central region) to visually search, browse, explore and discover multiple different movie titles of interest. The interface 20 may also be used to visually navigate other equivalent media items, sources or categories, such as books, music, apps, computer games, images, video footage, animations, etc.

[Example 2 - Constructing relationships between concepts

[0028] Contextual or semantic relationships between concepts and corresponding visual objects dynamically displayed in the interface of the invention are created, constructed or manipulated based on statistical or probabilistic methods, or mathematical algorithms, using meta data and data structures that reflect user interaction, history, feedback, and preferences. The relationships between concepts are represented numerically by relevance (or similarity) ratings maintained within a main data structure or memory table or map of a database, and the relevance ratings updated through user interaction with the interface. The aggregate result of all prior user interactions is used to return related concepts for subsequent use of the interface.

[0029] Figure 6 illustrates an example main data structure for dynamically constructing parent-child relationships between concepts represented by visual objects. Child data structures ('child a') for each concept are tracked within the main data structure. (Example data contained within each 'child a^* includes:

- storage of (or a reference to) a visual object representing the concept;

- a child data structure ('child b') for each relationship to another concept

- within each 'child b' the following data is specified: θ a numerical value (or additional child data structure) to rate (the relevance rating) the relationship between the child b concept and the child a concept;

a reference to (or method for accessing) the 'child a' version of the 'child b' concept;

- a reference or relevant data structure for dealing with a user interaction;

- optional data includes metadata or descriptive text to describe the concept

[0030] The main data structure illustrated in Figure 6 optionally also contains data and methods for evaluating the global properties of data contained within the children.

[0031] As described above, the relationships between concepts and corresponding visual objects are constructed through aggregate updating of the relevance rating using user relevance feedback. Information is communicated to the main data structure each time a user visually navigates between or interacts with concepts on the interface. The information includes where and how the user has chosen to navigate from a concept in the central region to a new concept or combination of concepts, and which concepts the user has interacted with. The information communicated to the main data structure is translated into operations that increase or decrease the value of the relevance ratings associated with the concepts on the interface at the time the user navigates or interacts by the interface.

[0032] In one example, the relevance ratings for related concepts are restricted to values between 0 and 1, and updated through user navigation and interaction accordingly. This is useful for constructing ratings for compound concepts, for example, by multiplying all of the 'child b' ratings for one component concept against all of the corresponding 'child b' ratings for the next component concept, and so on until all component concepts are accounted for. Other equivalent numbering systems may also be applied to the relevance ratings with appropriate adjustments for dealing with compound relationships.

[0033] The five main conceptual navigation operations provided by user interaction with visual objects in the interface are as follows. Simple navigation (as illustrated from Figure 2 to Figure 3) includes the following two actions:

(a) the relevance rating between the concept in the central region and the concept that the user navigated to is increased; and/or

(b) the relevance ratings between the concept in the central region and the concepts in the peripheral regions that the user did not select or navigate to are decreased.

Compounding navigation (as illustrated from Figure 2 to Figure 5) includes the following two actions:

(a) the relevance rating between the concept in the central region and the concept that the user dragged from the peripheral region into the central region is increased; and/or

(b) the relevance ratings between the concept in the central region and the concepts in the peripheral regions that the user did not navigate to are decreased.

Concept interaction (when a user interacts with a concept in a peripheral region surrounding the central region) includes the following two actions:

(a) the relevance rating between the concept in the central region and the concept that the user interacted with is increased; and/or

(b) the relevance ratings between the concept in the central region and the concepts in the peripheral regions that the user did not interact with are decreased.

Concept rejection (when a user rejects all related concepts in the peripheral regions by clicking on the central region) involves the single action of decreasing the relevance ratings between the concepts) in the central region and all of the concepts in the peripheral regions are decreased.

De-compounding navigation (where a user discards one of the concepts within the central region) includes the following two actions: (a) the relevance rating(s) between the discarded concept and the remaining concepts) within the central region are decreased; and/or (b) the relevance ratings between the discarded concept and the related concepts in the peripheral regions are decreased.

[0034] While the above operations describe increasing (decreasing) the relevance ratings to signify a stronger (weaker) relationship between two concepts, the opposite convention may also be applied.

[0035] When a user navigates to a new concept or compound concept, the new concepts returned to the peripheral regions of the interface are those that are considered most related, similar or relevant in terms of the value of their relevance rating. However, returning new related concepts to the peripheral regions of the interface presents a number of practical issues when considering the inherent use of the information presented by the interface. This. introduces four practical problems that it may be desirable to address before new related concepts are returned to the interface, or alternatively during updating of the relevance ratings.

[0036] The first problem is "lock-in" which is associated with consistently returning the most related concepts to the Interface. If merely the related concepts with the highest rating are returned to the peripheral regions of the interface when a given concept is introduced to the central region, then the same conceptual links will continue to be reinforced as the most related ones and the user is less likely to discover new concepts. New conceptual material will be prevented from appearing on the interface for consideration by a user. Likewise, potential evolution of conceptual relationships is hindered by the bias of the most related concepts receiving prominence. This problem is at least partially addressed by introducing a randomising engine that sits between the relevance ratings returned by the main data structure and the concepts delivered to the interface. With this solution, the relevance ratings do not just represent the strength of the relationship between two concepts, they are also probabilities (or biases in the randomising process) that the concept will appear on the interface when its counterpart is navigated to the central region. Most of the time the most related concepts will be delivered to the interface, however the act of randomising introduces the chance that loosely related concepts get the opportunity to appear on the interface. If a loosely related concept appears on the interface and is successful in enticing a user to interact with or navigate to it then it has a chance of becoming a more closely related concept. This process also allows other previously related concepts to be pushed out of consideration on the interface should their conceptual links become dated or decreasingly relevant.

[0037] The second problem concerns assigning "initial ratings" to concepts. When a new concept is added it may not have any definitive relevance ratings to other concepts within the main data structure. The initial ratings problem is at least partially addressed by indirectly updating the rating between potentially related concepts using "inheritance". When one of the concept navigation operations described above is applied, inheritance is applied to the relevance ratings of other concepts related to those concepts involved directly in the operation. Taking a simple navigation as an example, after the direct ratings between the concept in the central region and the one navigated to are increased, the relevance ratings between all concepts related to the concept in the central region and the concepts related to the navigated concept are jointly increased. The same procedure may be applied to the ratings of concepts that were updated in the initial inheritance operation and so on. Each level of inheritance may increase (decrease) relevance ratings at a finer magnitude than the last level. This process has the side effect of increasing the relevance ratings of marginally related concepts and giving them a chance to appear on the interface for direct consideration by users. A concept may eventually be deemed related to one or more concepts when given a chance on the interface, or its failure to result in interaction or navigation may see the concept return to obscurity once again. Giving a newly added concept some average initial relevance ratings to all or some of the concepts already represented in the main data structure may eventually see them reach the interface or become less and less relevant through indirect user feedback via the effects of inheritance. A second partial solution to the problem of initial ratings is the use, once again, of the randomising engine. This process helps concepts with small or average initial relevance ratings reach the interface for user consideration. Alternatively, newly added concepts are tagged specifically within the main data structure and they may be further biased for interface selection during the randomisation process.

[0038] The third problem is "genericism" which arises if a particular concept becomes repeatedly related to multiple other concepts so as to become a generic concept. Generic concepts are problematic because they hinder the ability to discover new concepts using the interface. The genericism problem is at least partially addressed by the introduction of a randomising process combined with a method or filter for identifying concepts that become generic. A method for identifying and filtering concepts approaching genericism is to track the total relevance ratings associated with one concept's relationship to all other concepts within the main data structure, such that the higher the total relevance ratings, the higher the level of genericism. Concepts with a high genericism rating can then be negatively biased within the randomising process.

[0039] The fourth problem concerns "duplicates" where two or more visual objects represent the same concept or closely related concepts. The duplicates problem also arises if two or more visual objects are considered to be identical or virtually identical based on past aggregate use of the interface. The duplicates problem is at least partially addressed by considering the respective relevance ratings that the two duplicate concepts respectively have with corresponding concepts. Both actual and virtual duplicates should yield similar patterns in the relevance ratings they each have with corresponding concepts. Firstly, the relevance rating between the two duplicates should be relatively weak, reflecting aggregate user rejection of the concepts yielding material difference between one another. Secondly, the relevance ratings that each concept has to other concepts should be quantitatively similar, reflecting the fact that a user who navigates to one of the concepts is essentially looking for the same type of content as users who navigate to the other concept. It is only when these two conditions are met that a duplicate can be considered to have occurred. Should the first condition not be met then this is indicative that users are interested in the relationship between the two concepts. Should the second condition not be met then this is indicative that each concept is either unrelated or virtually unrelated. The duplicates problem is further at least partially addressed by tracking potentially duplicate concepts within the main data structure and positively bias one of the duplicate concepts within the randomising process.

[0040] Other equivalent logic, data structures, statistical techniques, probabilistic methods, and mathematical algorithms may also be used to construct contextual, semantic and partially random relationships between selected and new concepts represented by visual objects on the interface in. ways that address or at least partially ameliorate one or more of the practical problems of lock-in, inheritance, genericism, and duplication referred to above.

Example 3 - Music: songs and albums

[0041] Figure 7 illustrates an example interface 20 for visually navigating music. Music has distinct hierarchical levels of interest, for example, on one hierarchical level users may be interested in exploring new individual songs, and on another hierarchical level they may be interested in exploring albums, compilations and soundtracks. In this example, the interface 20 provides different hierarchical levels for visually navigating between songs and albums.

[0042] Referring to Figure 7, a thumbnail of the song "A Day in the Life" by The Beatles is initially displayed in the central region 22 of the interface 20, and thumbnails of related songs are displayed in the peripheral regions 24. The interface 20 provides the same user navigations options described above in the movie example.

[0043] New songs are navigated to by clicking on one of the song thumbnails displayed in the peripheral regions 24. The selected song thumbnail is displayed in the central region 22, and thumbnails of related songs are displayed in the peripheral regions 24.

[0044] Alternatively, clicking on the central region 22 rejects all related song thumbnails in the peripheral regions 24. Thumbnails of new songs related to the original song are then displayed in the peripheral regions 24. Songs that are jointly related to Ά Day in the Life' and others songs represented by thumbnails in the peripheral regions 24 are selected by dragging one or more song thumbnails to the central region. Ail songs dragged into the central region 22 are visually represented by their corresponding thumbnails within the central region 22, and new related songs are displayed in peripheral regions 24.

[0045] In this example, the interface 20 provides a new option for navigating between hierarchical levels of songs and albums. By double-clicking on the central region 22, the interface 20 moves up a hierarchical level from songs to display related albums as illustrated in Figure 8. A user can now navigate and interact through albums in the same manner as previously navigating and interacting with individual songs.

[0046] Other equivalent interfaces 20 with multiple concept hierarchies may also be provided, and other equivalent user input may also be provided for navigating between different hierarchical levels. User interaction for moving up as well as down the hierarchy may be provided by, for example, double-clicking in the central region 22 to move down the hierarchy and double clicking in the peripheral regions 24 to move up the hierarchy.

[0047] Figure 9 illustrates a variant of the main data structure in Example 2 that has been adapted for containing relevance ratings associated with albums and individual songs within a hierarchy. Hierarchies of more than two levels are possible by extending the main data structure in the same manner used to extend the data structure illustrated in Figure 6 to the data structure illustrated in Figure Θ. For example, further hierarchies within the music application are provided by including artists as a level above albums. In this example individual songs are attributed to one or more albums, and each album is attributed to one or more artists (at the time of initial introduction to the application). The superscript "m" in the data structure illustrated in Figure 9 is not necessarily equal to the superscript V. The superscript "m" is limited by the number of objects contained within the object on the next hierarchical level above, using the example above this would be the songs contained within a particular album.

[0048] The introduction of concept hierarchies having different hierarchical levels creates two additional navigation operations to the interface 20 as follows.

1. Navigate up the hierarchy: the relevance ratings between the (lower hierarchical level) concepts) in the central region and the (lower hierarchical level) concepts in the peripheral regions are decreased.

2. Navigate down the hierarchy: the relevance ratings between the (higher hierarchical level) concept(s) in the central region and the (higher hierarchical level) concepts in the peripheral regions are decreased. [0049] The music example and the main data structure adapted for concept hierarchies illustrated in Figure 9 do not result in operations whereby a movement up or down the hierarchy alters or creates new ratings between individual songs and albums. Another further or alternative example implementation may allow for this within the data structure such that each 'child a' data structure contains additional children for ratings to each song within the population of songs. If this is implemented then the two further or alternative operations to those described above in paragraph [0048] may include one or both of the steps outlined below.

1. Navigate up the hierarchy: the relevance ratings between the (lower hierarchical level) concepts) in the central region and the (lower hierarchical level) concepts in the peripheral regions are decreased and the ratings between the (lower hierarchical level) concept(s) in the central region and the next selected or interacted (higher hierarchical level) concepts are increased.

2. Navigate down the hierarchy: the relevance ratings between the (higher hierarchical level) concept(s) in the central region and the (higher hierarchical level) concepts in the peripheral regions are decreased and the ratings between the (higher hierarchical level) concept(s) in the central region and the next selected or interacted (lower hierarchical level) concepts are increased.

[0050] The practical problems described above in paragraphs [0036] to [0040] are at least partially addressed in this music example by including a randomisation engine between the main data structure and the results returned to the Interface 20. The introduction of a hierarchical component to the interface 20 and main data structure reinforces the need to identify duplicates and virtual duplicates. For example, each individual song is attributable to an album, and each album contains individual songs so that duplicate songs that are contained within multiple albums will appear within the main data structure. The approach of identifying duplicates and favouring one in the randomisation process at least partially addresses the problem of duplicate songs. [0051] An additional problem of "hierarchical links" also emerges from the introduction of concept hierarchies in this music example; When a new hierarchical level is navigated to, new concepts should be relevant to the concept(s) that were residing in the central region 22 at the time of the hierarchical navigation. When moving up the hierarchy it makes sense to return the parent concept to the central region 22. The additional higher-level concepts in the peripheral regions 24 may be favoured in the randomising process via their relevance ratings to the parent concept in the central region 22, and the relevance ratings between their child concepts and the child concepts of the central region parent. When moving down the hierarchy, the concept(s) in the central region 22 may be all, some or one of the child concepts from the parentis) previously displayed in the central region 22 at the time of the hierarchical navigation. Furthermore, the child concepts within the peripheral regions 24 can be derived from a combination of the relevance ratings from the new child concepts) in the central region and the ratings of parent concepts prior to the hierarchical navigation. Identifying generic or duplicate children is again likely to be necessary, and in some embodiments they may each be used or excluded.

Example 4 - Travel planning

[0052] Figure 10 illustrates an example interface 20 for visually navigating travel planning. The interface 20, the main data structure and the concept of hierarchies are used to provide a visual travel planner for planning a trip to Italy. In this example, the central region 22 also provides a concept hierarchy as a cumulative travel itinerary that is built as the user navigates iteratively through concepts visually represented in the interface 20. This Itinerary is optionally displayed as a chronological list or a map within the central region 22.

[0053] An initial concept of Italy is displayed as a map or icon (illustrated as a frame with descriptive text for clarity) in the central region 22 based on a web search query, a suggestion based on web history or user profile, or a redirection from a web browser or web page. The interface 20 displays maps or icons of other potential destination countries (illustrated as frames with descriptive text for clarity) in peripheral regions 24. The interface 20 provides the same user navigations options described above in the movie example. For example, selecting and compounding two countries in the central region 22 of the interface 20 builds a travel plan involving the two selected countries.

[0054] Confirming a desire to travel to Italy via the interface 20, a user navigates down the hierarchy by double clicking on the central region 22 (or on the peripheral regions 24). The interface 20 provides photos or icons representing potential destinations within Italy as illustrated in Figure 11. The most popular, generic or representative Italian destination is initially displayed in the central region 22, and related destinations within Italy are displayed in the peripheral regions 24. Once again, the interface 20 provides a user with all the navigation options described above. The user optionally combines destinations and works out a broad travel plan before descending down the hierarchy and choosing accommodation and activities. Alternatively, the user populates a travel itinerary with accommodation and activities one destination at a time.

[0055] In the present example, a user chooses to explore a single destination at a time and decides to start in Rome. The user descends down the hierarchy via the interface 20 to select accommodation and is presented with the interface 20 illustrated in Figure 12. As illustrated in Figure 13, the user wishes to stay in more than one location in Rome, so the user decides to stay in the default suggestion in the central region, Hotel Majestic Roma, and adds Hotel Raphael by dragging the icon or photo representing Hotel Raphael into the central region 22. As illustrated in Figure 13, new hotel suggestions are now displayed in the peripheral regions 24 of the interface 20 that are jointly related to the two hotels the user has already selected in the central region 22. In this this example, the interface 20 has determined that the user wishes to navigate, browse and explore 5-star hotel accommodation in Rome.

[0056] Next, the user wishes to discover activities to do in Rome, so the user once again navigates down the concept hierarchy via the interface 20 and is provided with the interface 20 of icons or photos of attractions (illustrated as a frame with descriptive text for clarity) to visit illustrated in Figure 14. In this situation, the movement in the main data structure is more of a side step as the activities are things to do in Rome, not a subset of activities associated with the hotels. The user selects three activities illustrated in the central region 22 in Figure 15 and, based on the related activities displayed In the peripheral regions 24, the interface 20 infers that the user is predominantly interested in exploring ancient ruins and religious iconography.

[0057] At this point the user decides to complete planning of the Rome portion of the trip, and then ascends up the hierarchy vi¾ the interface 20 to select the next destination from Rome. Alternatively, as illustrated in Figure 16, the user views a summary of the selected itinerary as a hierarchical tree or list in the central region 22, and selects a hierarchical level of interest via the interface 20. Each level of the itinerary is a link that enables the user to revisit a particular level in the concept hierarchy. For example, by clicking on Rome the user is presented with destinations that are conceptually relevant to Rome - ie, they are near one another, have similar sites or for other reasons based on the aggregate interactions and navigations of Users who previously planned trips that included Rome. In the present example, once the user elects to return to the hierarchical level of destinations within Italy by selecting Rome in the itinerary displayed in the central region 22, the interface 20 reverts to that previously illustrated in Figure 11.

[0058] Having completed an itinerary for Rome, the user selects another destination within Italy to visit, and then uses the interface 20 to selectively fill out the details of the trip in the same manner as described for Rome. Alternatively, as illustrated in Figure 17 the user views the itinerary via the interface 20 using a map displayed in the central region 22 that indicates the user has decided to visit both Rome and Florence. In an extension to this example, the user optionally descends down the concept hierarchy via the interface 20 and selects restaurants that the user is interested in eating at whilst in Rome.

[0059] In this example, the only extension to the main data structure described above for the music example is the expanded use of hierarchies. Continued nesting of children as illustrated in Figure 9 is required when a new hierarchy is added to the application. Any additional operations beyond those outlined within the music example are generally unrelated to the main data structure. [0060] Issues associated with extending the visual interface 20 for an integrated travel planner are associated with setting defaults in terms of how new visual objects are displayed in the peripheral regions 24 as the user navigates through options on the interface 20. If a user decides to plan a trip in a top down manner (ie, the user first selects all of the countries that the user wishes to visit before descending down the hierarchy to map out the destinations in individual countries and so on) then a useful implementation is to default to a single parent concept rather than mixing child concepts at lower hierarchical levels. For example, assume the user decides to select Rome as the starting point, and then drags Florence into the central region 22 to map out the trip, before descending down the concept hierarchy via the interface 20. In this ease, the interface 20 initially defaults to hotels in Rome and indicates to the user that they must subsequently browse through hotels in Florence, and then activities etc.

Example 5 - Online grocery shopping

[0061] Figure 18 illustrates an example interface 20 for visually navigating online shopping. The interface 20 visually provides a dynamic shelf for a user to fill a shopping cart with selected grocery items. In this example, the user is able to visually navigate seamlessly through a variety of grocery items in a manner that is more intuitive and akin to the associative manner in which people think than simple shopping lists or physical supermarket aisles. The hierarchical extension described above is also necessary for this example.

[0062] The central region 22 of the interface 20 visually represents a shopping cart. The user drags thumbnails or icons of grocery Kerns into the central region 22 and the peripheral regions 24 displays thumbnails or icons of suggested new grocery items that the user may wish to purchase based on either relationships to the entire contents of the shopping cart, the last item dragged into it, or a combination of both. The interface 20 may infer (via the aggregate actions of past users) a higher-level concept of what the user wants and suggest or recommend items that help the user to complete or achieve that concept. For example, where a user selects eggs, butter and icing sugar, the interface 20 infers that the user is intending to make a cake and dynamically updates the peripheral regions 24 to display additional related ingredients such as flour, almond meal and cooking chocolate.

[0063] Figure 19 illustrates the interface 20 returning other grocery staples that the user may be interested in. The process 10 optionally gives the user specific control of when the suggestion engine runs. For example, if the user drags some grocery items into the shopping cart of the central region 22 - eggs, butter and coffee - rather than refreshing the grocery items in the peripheral regions 24 after one has been dragged into the central region, the interface 20 replaces the items dragged into the central region with others in the previous iteration that had just missed out on a place on the interface 20, or with new ones that are related to the combined contents of the shopping cart. This way the user does not miss an opportunity to add an item to the shopping cart. Further to this, if the user's opportunities stagnate, then the user can click in the central region 22 to indicate to the interface 20 to populate the peripheral regions 24 with new potential items. Figure 19 illustrates the case where the user has added items to the cart and only those items have been replaced. The interface 20 infers that the user is exploring or searching for breakfast items so that orange juice, bacon and baked beans are displayed in the peripheral regions 24 as suggested grocery items.

[0064] Thus far the user is using the dynamic "virtual" shelf provided by the interface 20 at a conceptual level. The hierarchical extension described above may also be used to enable the user to specify specific products via the interface 20. As illustrated in Figure 20, the user selectively descends down the concept hierarchy displayed in the central region 22 to specify particular products. Alternatively, at checkout the interface 20 prompts the user to select the specifics of what they want. If the user is in a rush, the interface 20 optionally automatically suggests and selects the most generic or popular individual product for the user. Similarly, at checkout the interface 20 optionally defaults to the most generic or commonly selected quantity of a particular grocery item, for example, a single loaf of bread, or five potatoes, etc. The interface therefore automatically populates the shopping cart of the user for checkout. [0065] Figures 21 and 22 illustrate implementation of the interface 20 for grocery shopping on mobile devices having small display screens, such as tablets, smartphones and mobile phones. To address the limited display screen size of such devices, the interface 20 allows the visual objects to be easily manipulated by reducing the size of the central region 22 so as to provide the peripheral regions 24 with more screen real estate on the device. The user is able to manipulate visual objects in the central region 22 by selectively zooming in and out to fill the available screen area. Figure 21 illustrates the process 10 with the interface 20 displaying the central region 22 inactive at normal size. Figure 22 illustrates the central region 22 activated and expanded to fill the available display screen area to enable user selection of magnified visual objects within the central region 22.

[0066] The dynamic and visual nature of the interface 20 makes it suited for small screen mobile devices to perform this grocery shopping example and the other examples described above and below. The mouse point, click and drag gestures described above may be implemented using equivalent touch screen gestures.

Example 6 - Online clothes shopping

[0067] Figure 23 illustrates an example interface 20 for visually navigating online clothes shopping. In this example the central region 22 displays an image of a person or mannequin, and images of clothing items are displayed in the peripheral regions 24. The interface 20 displays suggested clothing items in the peripheral regions 24 that are related to the selected clothing items displayed on the mannequin in the central region 22. The clothing items displayed in the peripheral regions 24 are updated dynamically as a user drags selected items onto the mannequin in the central region 22. The hierarchical extension of the process 10 described above may also be employed to enable a user to selectively navigate single types of clothing items in the peripheral regions 24 (ie, only business shirts or scarfs), or to drill down into more detailed customisations on a finished outfit, or to change all of the clothing items in the peripheral regions 24 to images of the central region 22 mannequin with the additional garment. [0068] The clothing example may be applied to equivalent examples involving customisable products or services, for example, furnishing a room, customising a car, customising a computer or home theatre, etc.

Example 7 - Online restaurant menus

[0069] In this example, the interface 20 is used as a visual restaurant locator to visually navigate online restaurant menus. The user presents the interface 20 with a restaurant that the user likes to eat at, and the interface 20 returns a collection of similar restaurants. The user is then able to browse through other restaurants via the interface 20 based on similar cuisines, location, price, customer ratings or some other attribute. This visual navigation of menu items is similar to the grocery example described above except that the available food options are limited to the contents of the menus of multiple selected restaurants. The interface 20 may also be used to customise meals from the collection of menus similar to the clothing example described above, for example, visually customising a pizza with user selected toppings. As with the movie and music examples described above, the interface 20 assists the user in visually discovering new restaurants and menu items that the user may never have discovered via a conventional keyword textual search. A mobile embodiment of this example may use the user's geo-location as the initial seed for returning, for example, the physically or geographically closest restaurant to the central region of the interface and related restaurants to the peripheral region.

Example 8 - Thesaurus. Phrases and Encyclopaedia

[0070] In this example, the interface 20 is used to visually navigate text. For example, the interface 20 may be configured as a visual thesaurus to explore related words. The interface 20 may also be configured to visually navigate and explore textual material such as phrases, quotes, blurbs, academic paper abstracts, etc. The additional conceptual complexity associated with academic paper abstracts provides potential use of the interface to infer higher-level concept hierarchies.

[0071] Another example is a visual language translator, for example, an English to Japanese translator where the visual objects displayed in the interface 20 are image or icons representing English phrases and Japanese counterparts. Visually navigating down the concept hierarchy allows a user to explore alternative translations in the event that there is no clear counterpart. The interface 20 may initially start with phrases and questions that are of a conversational nature, and the user may take a visual journey through the interface 20 to arrive at a level of language that is of a more technical nature. An initial directory Interface may be the most useful at directing the user to the type of language they are interested in, and this necessarily implies the use of the high-level conoepts hierarchies described above.

-Example 9 - General visual web browser

[0072] In this example, the interface 20 is implemented as a visual web browser or crawler having functionality generally similar to the examples described above. Further higher hierarchical conceptual extensions may be implemented through additional analysis of the aggregate ratings information that emerges from extensive use of the interface 20.

[0073] The visual objects used in this example are, for example, iconic visual representations of websites extracted by a web crawler as follows.

1. Infer an iconic image when crawling a given website by; grade all of the images displayed on the website by how linked they are to component and outside webpages; and then select the one image with the highest grade.

2. Take a snapshot image of a given website and display it as the visual representation.

3. Combine an iconic image with a zoom function to examine the snapshot.

[0074] Initial relevance ratings between websites are determined, for example, using a randomness process combined with inheritance as described above. There is also the added advantage that each website enters the main data structure with a set of initial links (or relevance ratings) to other potentially related websites. The interface is dynamically updated with new concepts using, for example, relevance feedback based on a metric that acts as a proxy or a measure for user satisfaction of what they derive from the interface. A simple metric is, for example, 'likelihood of an interaction with a new concept in the peripheral regions when a concept is initially presented to the interface.' This would keep all of the constant variables involved in a state of flux whereby they are continually adapted in a direction that appears to improve the value of the metric.

[0075] The main data structure used by the process 10 dynamically establishes new conceptual links between web pages. In this example, the interface 20 acts as a voting mechanism whereby the users (in aggregate) decide how they want the web to be linked from a conceptual point of view - it may be an indirect (and evolutionary) path towards the 'Semantic Web' that starts to develop links that users find to be more meaningful than the static links that were originally fed into the interface 20. Additional applications can then be developed that feed off of the meaning inferred from the conceptual links. Furthermore, a typical search engine may take into account the ratings of conceptual links such that more meaningful results are returned from search terms. The interface 20 has the potential to visually uncover broader meaning (it is not restricted to semantics), and different web pages (and the concepts within them) may be deemed relevant to one another for subjective reasons that exist in the aggregate consciousness of users. The users indirectly vote which relationships between which concepts are most meaningful to them. Furthermore, the interface itself encourages users to make new connections and gives them the option to connect seemingly unrelated or loosely related concepts. This mirrors the creative process and people may find the interface useful for visually discovering and even creating new ideas.

[0076] An example extension to the general web visual interface is an online news application where new concepts that have recently been crawled and added to the main data structure are given much higher weights in the randomising process than older material. This involves, for example, adding a timestamp to new concepts. The timestamp may be updated every time a user interacts with and navigates to a given concept. Further, the relevance ratings of concepts may be periodically degraded as their timestamps increase in age. This would have the added effect of restricting only new content to the interface 20. Eventually, a given concept may have all of its ratings degraded to a value of 0 and the concept may be expelled entirely from the main data structure.

[0077] Examples of the interface 20 may also be extended to finding contacts or people in online social networks such as Facebook, Linked In etc. In this example, the hierarchical extensions include, for example, job titles or organisational memberships, personal or professional networks, etc.

[0078] The examples described above have specified a version of the interface 20 where the relevance rating between two concepts A and B is symmetrical in that the rating for A -> B is equivalent to the rating for B => A (the implication restriction). A different version of the interface 20 allows these two ratings to vary from one another. This would simplify the updating operations as there would be no need to update the rating of a "child b's" corresponding 'child a' data structure. In addition, there would be a richer tapestry of information to work with in identifying different types of duplicate and generic concepts within the randomising process.

[0079] Embodiments of the present invention provide a new interface and user experience for visually navigating concepts on computing devices.

[0080] The above embodiments have been described by way of example only and modifications are possible within the scope of the claims that follow.

Claims

1. A method for visually navigating concepts on a computing device, including: displaying selectable visual objects representing concepts in central and peripheral regions of an interface rendered on a display of the computing device; receiving input representing one or more selections of one or more visual objects in the peripheral regions representing one or more selected concepts;

updating the peripheral regions of the interface to display new visual objects representing new concepts related to the one or more selected concepts.

2. A method according to claim 1 , further including receiving input representing one or more de-selections of one or more visual objects in the central or peripheral regions representing one or more deselected concepts, and updating the peripheral regions to display new visual objects representing new concepts related to the one or more selected concepts.

3. A method according to claim 1 , further including receiving input representing two or more selections of two or more visual objects in the peripheral regions representing two or more selected concepts, updating the central region to display the two or more selected visual objects, and updating the peripheral regions to display new visual objects representing new concepts having a compound relationship to the two or more selected concepts.

4. A method according to claim 3, wherein the two or more selected concepts are summarised in the central region as a tree structure or list representing a concept hierarchy.

5. A method according to claim 1 , wherein the new visual objects are displayed automatically based at least in part on relevance ratings between the new concepts and the selected concepts.

6. A method according to claim 5, wherein the new visual objects are displayed automatically based at least in part on randomising the relevance ratings between the new concepts and the selected concepts.

7. A method according to claim 1, wherein the central and peripheral regions of the interface are arranged in a two- or three-dimensional visual representation selected from a group comprising a matrix, a cluster, a grid, and an array.

8. A method according to claim 1, wherein the visual objects are interactive media items that visually represent content or physical items.

9. A method according to claim 8, wherein the interactive media items are thumbnails or icons.

10. A graphical user interface (GUI) for visually navigating concepts on a computing device, including:

displaying selectable visual objects representing concepts in central and peripheral regions of the GUI rendered on a display of the computing device;

receiving input representing one or more selections of one or more visual objects in the peripheral regions representing one or more selected concepts;

11. A GUI according to claim 10, further including receiving input representing one or more de-selections of one or more visual objects in the central or peripheral regions representing one or more deselected concepts, and updating the peripheral regions to display new visual objects representing new concepts related to the one or more selected concepts.

12. A GUI according to claim 0, further including receiving input representing two or more selections of two or more visual objects in the peripheral regions representing two or more selected concepts, updating the central region to display the two or more selected visual objects, and updating the peripheral regions to display new visual objects representing new concepts having a compound relationship to the two or more selected concepts.

13. A GUI according to claim 12, wherein the two or more selected concepts are summarised in the central region as a tree structure or list representing a concept hierarchy.

14. A GUI according to claim 10, wherein the new visual objects are displayed automatically based at least in part on relevance ratings between the new concepts and the selected concepts.

15. A GUI according to claim 14, wherein the new visual objects are displayed automatically based at least in part on randomising the relevance ratings between the new concepts and the selected concepts.

16. A GUI according to claim 10, wherein the central and peripheral regions of the GUI are arranged in a two- or three-dimensional visual representation selected from a group comprising a matrix, a cluster, a grid, and an array.

17. A GUI according to claim 10, wherein the visual objects are interactive media items that visually represent content or physical items.

18. A GUI according to claim 17, wherein the interactive media items are thumbnails or icons.

19. At least one computer-readable medium having stored thereon instructions for visually navigating concepts on a computing device, the instructions being executable to cause a computer processor to:

display selectable visual objects representing concepts in central and peripheral regions of an interface rendered on a display of the computing device; receive input representing one or more selections of one or more visual objects in the peripheral regions representing one or more selected concepts; update the central region of the interface to display the one or more selected visual objects; and