System and method for enlarging multi-media content

Abstract

The subject matter discloses a computer-implemented method for enlarging multi-media content without making any change to the multi-media content itself, the method comprising: creating a root of a tree; the root comprises the multi-media content; receiving a second multi- media content; the second multi-media content being retrieved by a search; the search utilizing 5 one or more keywords associated with a multi-media content object; the multi-media content object being identified by a tagging process conducted on the multi-media content; and creating child node of the tree; the node comprises the second multi-media content; and then repeating recursively the entire process to generate a tree of multi-media content that are all related to the enlarged multi-media content.

Description

[0001] SYSTEM AND METHOD FOR ENLARGING MULTI-MEDIA CONTENT

[0002] FIELD OF THE INVENTION

[0003] The present disclosure relates to multi-media content in general, and to video multimedia content, in particular.

[0004] BACKGROUND OF THE INVENTION

[0005] Video is more likely to go viral than other types of content because it’s easier to consume and understand. As of 2024, there are 3.9 billion videos on YouTube alone, of which there are 826 million Shorts. In 2023, 2.4 million videos were uploaded to YouTube every day. That equals 100,000 videos per hour, 1,750 per minute, and 29 per second.

[0006] Most videos on the Internet are designed to deliver a degree of limited knowledge, education, or entertainment to their viewers and are found on social media platforms such as Facebook or Instagram, video hosting platforms such as YouTube and Vimeo, and E-learning platforms such as Udemy and Coursera. Adding additional information to such videos beyond their existing content requires a solution such as editing the video content using tools like Adobe Premier Pro and enhancing the content where necessary, or simply completely re-constructing of the content of the video.

[0007] SUMMARY OF THE INVENTION

[0008] The term computing device refers herein to a device that includes a processing unit. Examples for such a device are a personal computer, a laptop, a server, a wearable device, a tablet, a cellular device, and IOT (internet of things) device.

[0009] The term multi-media content refers herein to any content that is stored on a computer- readable media. Examples of such content are a digital photo, a textual document, an audio clip, a blog post, a PDF document, a video clip or an animation (with or without audio), a webpage, a slide presentation, a virtual reality video, and the like.

[0010] The term media content object refers herein to an object of the multi-media content. Examples of such objects are an image within a photo or within a slide of a presentation, text in various formats and length on a video frame, a word within a subtitle of a narrator voice heard on a video, images or shapes or diagrams appearing on a slide presentation or video frames, an icon, a section of a video frame containing multiple images or shapes or diagrams or text or icons, Al generated content or any combination thereof.

[0011] The term knowledge-gap refers herein to a gap that may exist between the information exhibited by the multi-media content and the information an end user expects to or was intended to be provided by viewing the multi-media content. The term NLP refers herein to Natural Language Processing.

[0012] The term bubble refers herein to a structure that constructs a node in a tree for enlarging multi-media content. The terms bubble and node are used herein interchangeably.

[0013] The term meta-knowledge item refers herein to multi-media content of a bubble or a link to multi-media content of a bubble or parameters of a bubble associated with the multimedia content!

[0014] The term synergy refers herein to a certain degree of similarity between meta-knowledge items which creates, under certain pre-defined rule base, an association between bubbles.

[0015] The term similarity refers herein to a certain degree of resemblance between metaknowledge items.

[0016] The term proximity level refers herein to a degree of proximity calculated between the values of two parameters of a bubble associated with a multi-media content

[0017] The term presenting refers herein to displaying, playing or transmitting media to a computer device for displaying or playing.

[0018] The term end users’ behavior refers herein to the interaction of end users with the tree while watching a multi-media content. It identifies the number of visits in a node of the tree and / or in meta-knowledge items of a node, and the time and period of visits in a node of the tree and / or in meta-knowledge items of the tree.

[0019] The term computerized search refers herein to a search performed by a search engine on various data repositories and servers.

[0020] The term user refers herein to an end user or a content creator.

[0021] In some aspects of the present invention relates to a non - transitory computer - readable medium comprising instructions which when executed by at least one processor causes the processor to perform the method of the present invention.

[0022] One problem disclosed by the present disclosure is how to enrich an existing multi-media content with related content from outside source in order to address a knowledge-gap. Typically, multi-media content does not provide all the information that is relevant to the subject matter and the end user has to search for additional information on the Internet and switch from one search result to another in order to identify desired information. Such a problem may occur, for example, in short-form- videos which are deliberately made to be under one minute long in order to grasp the end user’s attention very quickly without including comprehensive information as may be expected by some video viewers. For example, a one-minute video intended to deliver basic knowledge about a medical procedure such as a Cataract operation, may focus on key areas of the procedure, yet in lieu of its short length, viewers who are seeking to understand the procedure in more depth, as well as its implications and risks, may find the information included on the content of the short video not sufficient to address their knowledge curiosity or needs. Viewers of such a medical video have to search for relevant medical knowledge elsewhere through search tools such as Google Search or Bing, and Al tools such as OpenAI’s ChatGPT™ or Google’s Gemini™. A process that consumes time and effort and not every simple end user has the skills to professionally sort through the mountains of search results and identify the correct information addressing his or her needs.

[0023] One solution disclosed by the present disclosure is embedding to the existing multi-media content, without any change to the existing multi-media content, a layer of a hierarchy of additional related multi-media content and presenting the additional multi-media content to the end user upon a query from the end user. In some embodiments, the additional multi-media content is of a subset of all possible combinations of relevant multi-media content and in accordance with criteria that is defined in the query. Examples of such criteria are popularity, similarity and synergy.

[0024] In one embodiment the layer is organized as a tree of data structures defined hereinafter as bubbles. Each bubble includes multi-media content or a link to this multi-media content and descriptive parameters of the multi-media content. Each bubble may also include keywords and links to additional and related multi-media content. In such a case the bubble is a parent node to other bubbles representing additional relevant multi-media content that is retrieved by a search that utilizes these keywords.

[0025] This structure enables to present additional relevant multi-media content in response to a query of the end user.

[0026] The additional relevant multi-media content may be retrieved from the internet or other data repositories manually by a content creator and / or with the assistance of Al (artificial intelligent) processing, and / or automatically using primarily Al processing.

[0027] Each bubble represents a node of a tree and is generated recursively.

[0028] The root node includes the original multi-media content.

[0029] Each of the other nodes includes additional related multi-media content that is retrieved by a search for enlarging the original multi-media content. The search is performed with keywords that are associated with the multi-media content of the parent node. In some embodiments the keywords are associated with a multi-media content object that is identified from the multi-media content of the parent node.

[0030] One exemplary embodiment of the disclosed subject matter is a computer- implemented method for enlarging multi-media content, the method comprising: creating a root of a tree for the multi-media content; receiving a second multi-media content; the second multimedia content being retrieved by a computerised search; the computerised search utilizing one or more keywords related to a media content object; the media content object being identified by tagging of the multi-media content; generating a child node of the tree; the child node comprises the second multi-media content; and linking the child node to the root to thereby enlarging the existing multi-media content with the second multi-media content. The method further comprising presenting the second multi-media content on a computing device of a user. The method, further comprising recursively repeating the generating the child node and the linking for generating one or more branches of the tree for enlarging each of the branches of the tree with enlarged multi-media content related to a parent of the branches. The method further comprising increasing a value of a counter of the child node, the counter being for counting visits of one or more end users in the child node and presenting the child node only if the counter being above a threshold. The method further comprising: receiving a similarity query from a user, the query comprises one or more reference meta-knowledge items; traversing the tree of the enlarged multi-media content or one or more trees including the tree of the enlarged multi-media content; identifying similar multi-media content complying with the similarity query, and presenting the similar multi-media content to the user. According to some embodiments the identifying of the similar multi-media content being if the reference meta-knowledge items of the similarity query being similar according to the similarity query to one or more second meta-knowledge items of a same type associated with the similar meta-knowledge items of the similarity query. The method further comprising receiving a synergy query from a user for presenting synergic multi-media content, the query comprises a reference multi-media content; traversing the tree of the enlarged multi-media content or one or more trees including the tree of the enlarged multi-media content; identifying synergic multi-media content, and presenting the synergic multi-media content to the end user. The method further comprising retrieving, from a data repository, a rule associated with synergy, the rule dictating a proximity level with related operands between an at least one meta-knowledge item of each multi-media content of a pair of nodes; identifying in a data repository a synergic group associated with the reference bubble; the synergic group comprises one or more synergic nodes; the each of the synergic nodes and the reference bubble comply with the rule; and presenting to the user the synergic nodes. According to some embodiments the synergic group being generated by a process of identifying proximity level between at least one meta-knowledge item of each pair of nodes in the tree and applying the synergy rule on each pair of nodes; whereby synergic nodes are defined as those that comply fully with the rule. The method further comprising applying a rule associated with past behavior of the end users with the tree pertaining to the number of visits of the multi-media content and its parameters, and the time of visits; extracting one or more multi-media content from the tree in accordance with the rule, the one or more multi-media content identifying trends in the end users’ behavior, and presenting the one or more multi-media content or the trends on a computing device.

[0031] One other exemplary embodiment of the disclosed subject matter is a computer- implemented method, the method comprising: receiving a query from a user for presenting synergic multi-media content, the query comprises a reference multi-media content; retrieving, from a data repository, a rule associated with synergy, the rule dictating a proximity level between an at least one meta-knowledge item of two bubbles in a tree; identifying in the tree a synergic group associated with the reference bubble; the synergic group comprises one or more synergic bubbles; each of the one or more synergic bubbles and the reference bubble comply with the rule; and presenting to the user the synergic bubbles. According to some embodiments the meta knowledge item being multimedia content associated with the bubble or a parameter describing the multimedia content. The method further comprising retrieving of a second multimedia content by a computerised search of one or more trees; the computerized search utilizes meta-knowledge items defined in a query.

[0032] One other exemplary embodiment of the disclosed subject matter is a system for creating a tree for enlarging multi-media content, the system comprising: one or more processors; and a memory communicatively coupled to the one or more processors, wherein the memory comprises instructions which, when executed by the one or more processors, cause the one or more processors to perform a method comprising: creating a root of a tree for the multimedia content; receiving a second multi-media content; the second multi-media content being retrieved by a computerised search; the computerised search utilizing one or more keywords related to a media content object; the media content object being identified by tagging of the multi-media content; generating a child node of the tree; the child node comprises the second multi-media content; and linking the child node to the root to thereby enlarging the existing multi-media content with the second multi-media content.

[0033] Embodiments of the invention may be implemented as a computer process (method), a computing system, or as an article of manufacture, such as a computer program product or a non- transitory computer-readable media. The computer program product may be a computer storage media readable by a computer system and encoding a computer program of instructions for executing a computer process on the computer and network devices. The computer program product may also be a propagated signal on a carrier readable by a computing system and encoding a computer program of instructions for executing a computer process. THE BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS

[0034] The present disclosed subject matter will be understood and appreciated more fully from the following detailed description taken in conjunction with the drawings in which corresponding or like numerals or characters indicate corresponding or like components. Unless indicated otherwise, the drawings provide exemplary embodiments or aspects of the disclosure and do not limit the scope of the disclosure. In the drawings:

[0035] Fig. 1 shows a block diagram of a system for enlarging multi-media content, in accordance with some exemplary embodiments of the subject matter;

[0036] Fig. 2 shows a block diagram of a bubble structure, in accordance with some exemplary embodiments of the disclosed subject matter;

[0037] Fig. 3 shows a flowchart diagram of a scenario for generating a bubbles’ tree for enlarging multi-media content, in accordance with some exemplary embodiments of the disclosed subject matter;

[0038] Fig. 4 shows a flowchart diagram of a process for extracting recommended multi-media content from the tree, in accordance with some exemplary embodiments of the disclosed subject matter;

[0039] Fig. 5 shows a flowchart diagram of a process for extracting similar multi-media content from the tree, in accordance with some exemplary embodiments of the disclosed subject matter

[0040] Fig. 6 shows an exemplary user dashboard, in accordance with some exemplary embodiments of the disclosed subject matter;

[0041] Fig. 7 shows an exemplary bubbles’ tree, in accordance with some exemplary embodiments of the disclosed subject matter;

[0042] Fig. 8 shows a flowchart diagram of a process for identifying synergy among bubbles, in accordance with some exemplary embodiments of the disclosed subject matter;

[0043] Figure 9 shows an exemplary proximity matrix, in accordance with some exemplary embodiments of the disclosed subject matter;

[0044] Figure 10 shows exemplary synergic groups of bubbles, in accordance with some exemplary embodiments of the disclosed subject matter; and

[0045] Figure 11 shows a flowchart diagram of a Learn and Deduct process, in accordance with some embodiments of the disclosed subject matter.

[0046] DETAILED DESCRIPTION

[0047] Fig. 1 shows a block diagram of a system for enlarging multi-media content, in accordance with some exemplary embodiments of the subject matter. System 100 includes a module for creating bubbles 101, a module for traversing through bubbles 102, a dashboard module 103, a synergy module 104, a similarity module 105, a proximity module 106, rule base module 107, Learn and Deduct module 108 and data repository 109.

[0048] The module for creating bubbles 101 is configured for creating a tree of bubbles for enlarging a certain multi-media content. The operation of the module for creating bubbles 101 is explained in greater details in figure 3.

[0049] The module for traversing through bubbles 102 is configured for traversing through the bubbles in the tree for performing knowledge mining. The knowledge mining is for exploring and analyzing the content of the bubbles in the tree. The traversing can be performed by methods such as Level-Order Traversal. The traversing may be performed on the whole tree or on a certain branch containing only a portion of the tree.

[0050] The dashboard module 103 is configured for interacting with the end user. The interaction with the end user is for accepting knowledge mining requests, processing such requests, and then presenting the results of such requests to the end user. The operation of the dashboard 103 is explained in greater details in figure 6.

[0051] The synergy module 104 is configured for identifying synergy between bubbles. The operation of the synergy module 104 is explained in greater details in figure 8.

[0052] The similarity module 105 is configured for identifying similarity between bubbles in accordance with a query. The operation of the similarity module 105 is explained in greater details in figure 5.

[0053] The proximity module 106 is configured for identifying proximity between metaknowledge items contained within bubbles. The proximities are identified by comparing all combinations of pairs of meta-knowledge items of the same type. For example, the title of bubble(i) and the title of bubble(j) in the pair of bubbles (i,j) are compared to each other and may have a proximity level of <high>, or the descriptor of bubble(i) and descriptor of bubble(j) in the pair of bubbles (i,j) are compared to each other and may have a proximity level of <medium>, and so on. The proximity module 106 is utilized by the synergy module 104 for identifying the synergy between bubbles.

[0054] The rule base module 107 is configured for identifying rules. Some rules are for identifying synergy between bubbles and some other rules are for identifying recommendations and performing analytics by the Learn and Deduct process.

[0055] The Learn and Deduct module 108 is configured for performing the Learn and Deduct process. The Learn and Deduct process is for collecting end-users’ behavior during knowledge mining, identifying recommended multi-media content from collected behavior, calculating analytics for estimating trends in end users’ viewing of multi-media content, and identifying required improvements in the bubbles’ tree given trends. The Learn and Deduct process is explained in greater details in figure 11.

[0056] The data repository 109 is configured for storing the bubbles’ tree, rules, configuration parameters, end users’ behavior, recommendations generated by the Learn and Deduct process, and synergy data generated by the synergy process. The modules communicate with the data repository 109 for retrieving data and for updating the data repository.

[0057] Fig. 2 shows a block diagram of a bubble, in accordance with some exemplary embodiments of the disclosed subject matter.

[0058] Bubble 200 includes meta-knowledge items. The meta-knowledge items include title 201, type 2022category 203, keywords 204, counter 205, descriptor 206, link to multi-media content 207, link to parent bubble 208 and end-users vector 213.

[0059] The title 201 includes the title of the multi-media content of 207.

[0060] The type 202 includes the type of the multi-media content of 207, which may be, for example, video, image, webpage, document, or Al-generated content, and audio.

[0061] The category 203 includes category of the multi-media content of 207, for example if the category is Medicine the multi-media content of 207 contains content associated with such category. There may be a plurality of categories for each bubble, and in such a case the category may be a vector.

[0062] The keywords 204 includes keywords for searching the additional multi-media content and a link to the additional multi-media content associated with a search conducted with each keyword. In some cases, the content creator does not insert any keywords and the bubble represents a leaf node in the tree.

[0063] For illustration purposes only, keyword A is connected to child A bubble 211 and keyword N is connected to child N bubble 212.

[0064] The counter 205 includes a counter of the total number of end users that visited bubble 200.

[0065] The end-users vector 213 is configured for storing the history of visits of end users. The end-user vector 213 can be, for example, in the following format: {Bubble, [(end user, metaknowledge item, date visited, time- visited), (end user, meta-knowledge item, date visited, time- visited), (end user, meta-knowledge item, date visited, time-visited)... ] }

[0066] The descriptor 206 includes the short description of the multi-media content 207.

[0067] The link to multi-media content 207 includes a link to the content or a copy of the content for enlarging the multi-media content of the parent bubble 210.

[0068] The link to the parent bubble 208 connects the bubble to the parent bubble 210. Fig. 3 shows a flowchart diagram of a scenario for generating a tree for enlarging multimedia content, in accordance with some exemplary embodiments of the disclosed subject matter.

[0069] At block 300 the multi-media content that needs to be enlarged is uploaded to the platform workspace. An example of such multi-media content is a video clip in the form of .MP4, .AVI, .MOV, .WMV file. The multi-media content can be uploaded, for example, from a video sharing platform (e.g., YouTube, Vimeo), or an e-learning platform (e.g., Coursera, Udemy, edX) or a social networking platform (e.g., Facebook, Linkedln, TikTok) or a server, and the like.

[0070] In one example the multi-media content is a video clip related to cataract surgery.

[0071] At block 305 the root of the bubbles’ tree is generated. The root points to the multi-media content that needs to be enlarged. The root is in the form of a bubble. The structure of the bubble is explained in greater details in figure 2.

[0072] At block 310 the system identifies if a condition for terminating the creation of the tree is met. The condition refers to a maximum number of levels that can be formed in the tree.

[0073] The maximum number of levels of the tree is pre-defined by the content creator or automatically by the system as a default value.

[0074] If the maximum number of levels of the tree is reached and if there is no more requirement for enlarging the tree with additional bubbles, then the operation of creating the tree terminates at block 315.

[0075] Otherwise, at block 320 the multi-media content is tagged. The tagging is for identifying media content objects within the multi-media content. The tagging may be done automatically by Al auto-tagging methods such as Visual Intelligence™ of AnyClip.com. Al auto-tagging is the process in which artificial intelligence is used to automatically tag media files with metadata through visual recognition of their content. This is an approach to automate tagging, which creates a term that describes a keyword or phrase according to a visually recognized content and assigns such term to a tag for such visually recognized content. For example, on an animal video, a media content object of a dog image of the type of Bulldog may be recognized by Al autotagging and may be assigned the term” dog”.

[0076] In some embodiments the tagging may be done manually by the content creator by identifying media content objects on the multi-media content and manually tagging such objects. Al auto tagging and manual tagging can be done side by side. For example, if the enlarged multimedia content is a video about home pets, the Al auto tagging process for such video may recognize the images of a Siamese cat or a Bulldog on certain frames of the video and create a tag for each of these images automatically as well as assign related terms to such tags. The content creator can then manually recognize on the same video a bowl with pet food that was not recognized by the Al auto tagging process and add manually a tag and a term for the bowl.

[0077] In both Al auto tagging and manual tagging, the tagging operation may also assign some metadata values such as the time stamp of the tag from the beginning of the multi-media content (e.g., in case of a video clip,) and the {x,y} coordinates of the tag for the media content object within a frame in case of a video clip or a slide in case of a slide presentation or a digital photo.

[0078] In some embodiments the enlarged multi-media content may be three dimensional and tagging shall be done with a positioning of a tag in the {x,y,z} coordinates. For example, in the case of a virtual reality video, the position of tags may be done in this way and the corresponding metadata for the coordinates of the tag shall be {y,y,z}.

[0079] In some embodiments, the system operates a speech-to-text process for generating transcript of voice and then tagging such transcript using Al methods such as NLP that are able to identify accurately meaningful words within text or Google’s Gemini™ Al-powered assistant which is able to accept a prompt such as “generate keywords for text: <here comes the text>” and generate a response with a list of meaningful keywords. For example, if the enlarged multimedia content is a video about animals and it includes a narrator voice, a transcript of such narrator voice can be generated by using a tool such as Google Cloud Speech API to convert the narrator voice to transcript. The transcript of the voice is then used for identifying meaningful keywords or phrases using Al methods such as NLP or tools such as Google’s Gemini™ AI- powered assistant. Such meaningful words or phrases become tags for media content objects of the transcript. For example, if the transcript of a video about animals has the word “dog”, and an Al method such as NLP recognized such word, then the word shall be tagged with a special color, and the term “dog” shall be assigned to the tag. In some embodiments the content creator can manually tag a transcript. In both Al tagging and manual tagging of transcript, tagging also assigns some metadata values to each tag of a word or a phrase such as the time stamp of the word from the beginning of the voice of the transcript (e.g., in case of a video clip).

[0080] If the tagging is done by Al tools such as Visual Intelligence™ in the case of frames, or an Al method such as NLP or a tool such as Google’s Gemini™ Al-powered assistant in the case of transcript, the content creator may select one or more media content objects from the results of the tagging for which bubbles are to be created.

[0081] For example, the media content objects associated with cataract surgery are retina and pupil. If the Al auto-tagging process creates a tag for these media content objects, the content creator may select both retina and pupil as media content objects for which a bubble shall be created. In some embodiments the system may select automatically one or more media content objects from the results of the tagging for which bubbles are to be created. For example, such selection may be done if the term of the tag is popular. This can be determined by similarity comparison of the term with the meta-knowledge item title for all bubbles in the tree, and according to the result of the similarity comparison and the end users’ behavior infer if the term is popular, hence it should be selected for new bubble.

[0082] Blocks 325 and 330 are performed per each selected media content object for identifying search parameters and creating bubbles. The parameters are used for performing a search for retrieving additional multi-media content that is related or relevant to the current media content object.

[0083] At block 325 the content creator assigns keywords associated with the media content object via a dashboard.

[0084] In some embodiments the system may assign automatically some or all of the keywords by using tools such as chatGPT™ or Gemini™ to create a generative Al prompt from the term of the tag and any additional relevant metadata provided by the Al auto-tagging process to describe the tag (e.g., a category), generating a response, analyzing the response with Al methods such as NLP to identify meaningful words, and using such meaningful words as keywords.

[0085] In some embodiments keywords are identified by the content creator by using generative Al tools such as chatGPT™ or Gemini™ to create a response to generative Al prompt and applying an Al method such as NLP on the response to the prompt to identify keywords within such response.

[0086] For example, in the case of a cataract explainer video, the chatGPT query “what are the parts of the retina” may produce the following response “The retina is made of two parts, the macula and the peripheral retina. The macula is in the center of your retina and processes most of what you're directly looking at. The peripheral retina fills in the parts of your vision at the edges of your visual field (your peripheral vision)”. An Al method such as NLP may identify several keywords from within the response such as “macula” and “peripheral retina” and “peripheral vision”.

[0087] The keywords are used for retrieving additional multi-media content that is related to the media content object.

[0088] At block 330 the content creator assigns parameters for the search via a dashboard. The parameters are related to the media content object, and include, inter aliya, title, category (one or more), a type of the desired multi-media content (e.g., video, audio, document, generative Al information, or any combination), the keywords identified on 325, and optionally a text descriptor of the media-content object. In some embodiments some or all the parameters for the search are assigned by the system automatically.

[0089] Blocks 335, 340 and 345 are performed per each keyword.

[0090] At block 335 the system performs a search utilizing the keywords and the parameters of the search for retrieving additional multi-media content that is related or relevant to the current media content object. The search may be performed on the internet or in a local database or a local server or a combination thereof. For example, the search may find on the Internet one or more videos or one or more web-pages associated with peripheral retina. The search may be performed by activating tools such as Google Search or a proprietary search engine. The system may select from the search results the most relevant multi-media content. The most relevant multi-media content may be chosen by utilizing a search engine prioritization process or a proprietary prioritization process. In some cases, the content creator selects manually the most relevant multi-media content from all search results. In some embodiments the search may be done on a reservoir of one or more other bubbles’ trees generated by one or more content creators for enlarging other multi-media content.

[0091] At block 340 the system generates a bubble for the selected search result. The bubble includes the meta-knowledge items identified on the search. The meta-knowledge items include parameters such as the title of the bubble, the type of the multi-media content, one or more categories, and additional parameters such as a descriptor of the multi-media content associated with the selected search result. In some embodiments, some of the meta-knowledge items may be retrieved automatically by the system from the metadata associated with the multi-media content of the bubble. For example, if the multi-media content is a video uploaded from YouTube or other video hosting platform, such video may also be accompanied by various metadata parameters defined by the video creator while the video was uploaded to the YouTube platform. These metadata parameters may include a title, a category, one or more keywords, and a descriptor of the content of the video.

[0092] At block 345 the system links the new bubble to its parent bubble. The parent bubble is the bubble that includes the multi-media-content from which the media content object is identified and tagged. The link is for generating a tree of bubbles that expands a knowledge base associated with the multi-media content of the root 305. The recursive operation resumes to block 310 for recursively generating the tree.

[0093] In some embodiments the system analyzes the behavior periodically, and according to some inference rules, it infers if there is a need to enhance the tree with more bubbles. If so, the system creates such bubbles given the inference rule. For example, the inference rule can be, “check if the bubble visit counter is greater than 25% above the mean of all visit counters of all bubbles in the tree'. If the result of the check is positive, then the system generates a new branch for the bubble using the following method: engineer a generative Al prompt from the bubble parameters such as title and categories; generate a response for the prompt; use an NLP method to identify keywords in the response, or use a tool like Google’s Gemini™ with a prompt such as: "list keywords from text: <response>" to identify keywords in the response; then use such keywords for a new search and proceed to create a branch of bubbles for each keyword. This process can then continue to one or more levels recursively.

[0094] Fig. 4 shows a flowchart diagram of a process for retrieving recommended multi-media content from the tree, in accordance with some exemplary embodiments of the disclosed subject matter.

[0095] According to some embodiments the system presents the user, upon request, the option to view recommended content. The recommended content is defined according to a counter of the number of visits of the bubble. If the counter exceeds or equals to a pre-defined threshold and / or exceeds or equals to a pre-defined percent above the calculated mean of all counters of visits in all bubbles, the bubble is defined as recommended. For example if there are 10 bubbles in the tree with the following counters {200, 53, 5, 150, 60, 1, 140, 75, 0, 86}, the system sorts the list of bubbles by their counters {200, 150, 140, 86, 75, 60, 53, 5, 1, 0}; then the system calculates the percentages of each counter above their mean of 77 to be: { 160%, 95%, 82%, 12%, -3%, - 22%, -31%, -94%, -99%, -100%}; given the calculation, the system may set the bubbles with the counters {200, 150, 140, 86} as recommended.

[0096] The process is performed by Learn and Deduct process.

[0097] Referring now to the drawing:

[0098] At block 400 the system initiates the tree traversing operation. The traversing can be performed by methods such as Level- Order Traversal.

[0099] Blocks 405, 410 and 415 are performed per each node.

[0100] At block 405 the system checks if the counter of the number of visits associated with the node is above a pre-defined threshold and / or exceeds or equals to a pre-defined percent above the calculated mean of all counters of all other visits in other nodes.

[0101] If so, then at block 410 the system copies the node to the chain of recommended nodes; otherwise, the system continues to block 415.

[0102] At block 415 the system checks if the current node is the last one.

[0103] If so, then at block 420 the process is terminated, and the chain of recommended nodes is completed.

[0104] Otherwise, at block 425 the system moves to the next node. Given the chain of recommended nodes, the recommended multi-media content of these nodes can be presented to the user upon request by traversing the nodes of the chain and presenting the multi-media content that is associate with each node.

[0105] Fig. 5 shows a flowchart diagram of a process for extracting similar multi-media content from the tree, in accordance with some exemplary embodiments of the disclosed subject matter.

[0106] According to some embodiments the user may query the system for viewing similar multi-media content according to similarity between meta-knowledge items included in the query and meta-knowledge items of the same type included in compared bubbles in the tree as well as operands such <and>, <or>, <less than percent similarity> and <greater than percent similarity >. In some embodiments the similarity check applies to bubbles in multiple trees stored in a reservoir of one or more trees.

[0107] Initially, the user issues a query that includes parameters for identifying the required similarity and if needed some operands. Examples of such parameters are the type of the multimedia content (e.g., video, document, Al-generated, audio etc.), one or more keyword(s), a descriptor, and one or more categories. For example a similarity query may consist of the following: { titlc=”hcart”, typc=”all”, key word=” valve”, similarity>=80%”}. Such query instructs the system to identify bubbles with similarity of at least 80% for all parameters included in the query when compared with the parameters of the bubbles. The system then traverses the tree for identifying similarities according to the query and extracting the similar multi-media content for the user. Referring now to the drawings:

[0108] At block 500 the system receives the query from the user. The query includes any combination of parameters such as type of multi-media content sought after (video, webpage, image, document, audio etc.), one or more keyword(s), a descriptor, and one or more categories. In some embodiments the query may be formatted as free text. In such a case the system may use an Al method such as NLP for extracting the parameters of the query. An example of such query may be: “Please provide me all the videos that are associated with the words cat and or dog”. The NLP extracts the words “video”, “cat” and “dog”, and the query syntax is constructed from the free text as follows: | typc=VIDEO. keyword=”cat”, key word=”dog” } . In some embodiments the system may recommend to the user one or more relevant queries to choose from using a reservoir of queries that other users have composed in the past to identify similarities in the tree.

[0109] At block 505 the system starts the traversing of the tree. The traversing can be performed by methods such as Level-Order Traversal.

[0110] Blocks 510, 515 and 520 are performed per each node.

[0111] At block 510 the system tests if the parameters of the query match the parameters of the bubble representing the compared node according to the definition of the query. For example, for the query “Please provide me all the videos that are associated with the words cat or dog”, the system tests if the compared node is associated with a video clip and includes in at least one of its parameters, such as the title, the keywords, or the descriptor, or any of its categories, the string “cat”. If the answer for such test is true than there is a match. In some embodiments the system may also infer that the words “dog” or “cat” are associated with the category “animals”, and if such a category is set for the compared node, the system concludes there is a match.

[0112] If there is a match, then at block 515 the system copies the compared node to the chain of nodes representing a similarity to the query; otherwise, the system continues to block 520.

[0113] At block 520 the system checks if the compared node is the last one.

[0114] If so, then at block 525 the system presents the results in the chain to the user and terminates the process, otherwise at block 530 the system moves to the next node and the comparison operation resumes to block 510.

[0115] It should be noted that, at some cases the user may request to choose the relevant content out of recommended nodes (bubbles). For example, the user may create the query “all recommended bubbles with the word dog”. In such a case the system traversers the recommended nodes in the tree only instead of the entire tree and look for matching bubbles that are recommended and also contains in one of their parameters the word “dog”.

[0116] Fig. 6 shows an exemplary dashboard, in accordance with some exemplary embodiments of the disclosed subject matter.

[0117] Dashboard 600 is configured for enabling the user to perform knowledge mining and view desired multi-media content.

[0118] Dashboard 600 displays a video frame 601. The video frame 601 includes tagged media content objects 6011.

[0119] Dashboard 600 includes navigation buttons 602. The navigation buttons are for navigating through the tree of bubbles or a chain of bubbles that comply with a certain criterion such as recommended bubbles and reaching desired multi-media content.

[0120] Dashboard 600 includes knowledge mining operations 603 for including manual traversing of the tree of bubbles, generating lists of recommended multi-media content, performing similarity queries, and generating various synergies.

[0121] Dashboard 600 includes control buttons 604 for controlling the dashboard operation.

[0122] Fig. 7 shows an exemplary tree, in accordance with some exemplary embodiments of the disclosed subject matter.

[0123] The root node of the tree includes a video of cataract surgery 70.

[0124] The root node of the tree includes 2 branches. The left branch includes multi-media content related to retina and the right branch includes multimedia content related to pupil. The root of the left branch is a node with multi-media content related to retina 71.

[0125] Node 71 points to a node with multi-media content related to peripheral retina 711, a node with multi-media content related to macula 712, and a node with multi-media content related to fovea 713.

[0126] Node 712 points to a node with multi-media content related to pigmentation 714.

[0127] The root of the right branch is a node with multi-media content related to pupil 72 which points to a node with multi-media content related to iris 721.

[0128] Fig. 8 shows a flowchart diagram of a process for identifying synergy among bubbles in a tree in accordance with some exemplary embodiments of the disclosed subject matter. In some embodiments the system identifies synergic multi-media content among one or more trees. The synergy among multi-media content is identified according to predefined rules associated with proximity levels between pairs of same type meta-knowledge items from two bubbles.

[0129] An example of such a rule applied on pair of meta-knowledge items in bubbles (i,j) is: {HIGH proximity level for title OR MEDIUM proximity level for descriptor. } For such a rule the system performs comparisons of all combinations of pairs of bubbles in the tree, and for each such pair compares the titles to each other, and check if such comparison falls within a proximity level of <high> ; then the system performs comparisons of all combinations of pairs of bubbles in the tree, and for each such pair compares the descriptors to each other, and check if such comparison falls within a proximity level of <medium>. The system then selects the pair of bubbles that comply with the rule; that is to say the pairs (i,j) of bubble(i) and bubble(j) that have <high> level of proximity of titles or <medium> level of proximity of description.

[0130] In some embodiments, the synergy may be identified among recommended bubbles of the tree. In such a case the system traversers the recommended nodes only instead of the entire tree.

[0131] Referring now to the drawing:

[0132] At block 800 the content creator defines proximity levels. Examples of proximity levels are <weak> for up to 40% proximity, <medium> for 41% proximity up to 75% proximity, <high> for 76% proximity to 99% proximity and <exact> for 100% proximity.

[0133] At block 805 the content creator defines one or more synergy rules for identifying synergy between a pair of bubbles. The rules include definition of proximity levels for one or more types of meta-knowledge items associated with a pair of bubbles with operands such as AND, and OR.

[0134] Such a rule may be, for example: {HIGH proximity level for title and MEDIUM proximity level for descriptor}.

[0135] Blocks 810, 815, 820, 825, 830 and 835 are performed per each defined rule. At block 810 the system compares each of the meta-knowledge items that are included on the inclusion column 903 as depicted in Figure. 9, for all the combinations of pairs of nodes in the tree. The comparing is for finding proximity levels between the included meta-knowledge items for all the combinations of pairs of nodes.

[0136] For example, if the tree includes bubbles 1, 2 and 3 and the inclusion column 903 of figure 9 includes the title and the descriptor, the system performs the following:

[0137] Comparing the title of bubble 1 to the title of bubble 2 and identifying the proximity level, comparing the title of bubble 1 to the title of bubble 3 and identifying the proximity level, and comparing the title of bubble 3 to the title of bubble 2 and identifying the proximity level.

[0138] Then, comparing the descriptor of bubble 1 to the descriptor of bubble 2 and identifying the proximity level, comparing the descriptor of bubble 1 to the descriptor of bubble 3 and identifying the proximity level, and comparing the descriptor of bubble 3 to the descriptor of bubble 2 and identifying the proximity level.

[0139] At block 815 the system stores the results of the comparisons in a proximity matrix. The proximity matrix identifies the pairs (i,j) of bubble (i) and bubble (j) that comply with each proximity level per each meta-knowledge item in the inclusion column 903 of figure 9. An example of such proximity matrix is depicted and explained in figure 9.

[0140] At block 820 the system applies each synergy rule on all pairs in the proximity matrix, and then the system retrieves from the proximity matrix, the pairs of bubbles that comply with each synergy rule.

[0141] For example, if the rule is {MEDIUM proximity level for title AND EXACT proximity level for type}, the system retrieves from the matrix all the pairs that have both <medium> proximity of their titles and <exact> proximity of their types.

[0142] If, for example, the rule is {MEDIUM proximity level for title OR EXACT proximity level for type}, then the system retrieves from the matrix all the pairs that have either <medium> proximity of their titles or <exact> proximity of their types. The system inserts each such pairs into a vector of synergic bubbles associated with the current rule.

[0143] At block 825 the system generates a synergic group per each bubble in accordance with the data of the vector of synergic bubbles. Each group includes a certain bubble and all other bubbles that are synergic to this bubble. For example, if bubble 1 is synergic with bubbles 2 and 5, then the system generates the group 1,2 and 5. An example of such synergic groups is depicted in figure 10.

[0144] At block 830, the user issues a request for retrieving all the bubbles that are synergic to a chosen reference bubble. For example, the user requests the bubbles that are synergic to bubble 3. At block 835 the system presents to the user the multi-media content of all bubbles that are synergic to the chosen reference bubble.

[0145] It should be noted that in some embodiments the user may request to choose the synergic content out of recommended content. In such a case the system may traverse and perform synergy processing on the recommended nodes in the tree instead of the whole tree.

[0146] Figure 9 shows an exemplary proximity matrix, in accordance with some exemplary embodiments of the disclosed subject matter.

[0147] The columns 901 of the proximity matrix 900 include proximity levels between metaknowledge items of the same type associated with two bubbles. The proximity levels are <exact>, <high> and <medium>.

[0148] The inclusion column 903 defines if the meta-knowledge items are to be included in the comparison. For example, if all parameters are to be included than they are marked positively for comparison. The meta-knowledge items that are not marked in the inclusion column 903 are excluded from the comparison.

[0149] The lines 902 of the proximity matrix 900 include the definition of the meta-knowledge items for comparison. In some embodiments the multi-media content can also be included in the comparison though not shown on the figure.

[0150] A cell in the matrix includes the pair of bubbles {i,j } , bubble(i) and bubble(j), that have the proximity level represented by the column of this cell for the meta-knowledge item that is defined by the line of this cell. For example, the cell in the proximity matrix for the column with <high> proximity level and the line of the <title> parameter contains the pair (4,8) representing bubble 4 and bubble 8 compared for the <title> parameter. The cell in the proximity matrix for the column with <high> proximity level and the line of the Category parameter may have the same pair (4,8) representing bubble 4 and bubble 8 compared for one of the categories of the Category parameter. The cell in the proximity matrix for the column with <medium> proximity level and the line of the <Keyword> parameter may have the same pair (4,8) representing bubble 4 and bubble 8 compared for one of the keywords of the <Keyword> parameter.

[0151] If the rule {HIGH proximity level for title AND HIGH proximity level for category AND MEDIUM proximity level for keyword} exists in the pre-defined rule base, then when it is applied on the proximity matrix 900, Bubble 4 and Bubble 8 shall be identified by the system as synergetic to each other as the combination of comparisons of their parameters comply with the rule. It should be noted that at some cases if a rule includes a proximity level such as <high> it also covers the proximity level of <medium>, i.e., any rule that includes a proximity level that is higher covers pairs that have a lower proximity level on the matrix. This can be exhibited in the case of category and keyword parameter which may include multiple different categories and multiple different keywords.

[0152] Figure 10 shows exemplary synergic groups of bubbles, in accordance with some exemplary embodiments of the disclosed subject matter.

[0153] The synergic graph 10 depicts the synergy among nine bubbles.

[0154] Bubble 1 (11) is synergic with bubble 2(12) and bubble 3(13).

[0155] Bubble 2 (12) is synergic with bubble 1(11), bubble 4(14) and bubble 9 (19).

[0156] Bubble 3 (13) is synergic with bubble 1(11) and bubble 5(15).

[0157] Bubble 4 (14) is synergic with bubble 2(12), bubbles 5 (15) and bubble 6 (16).

[0158] Bubble 5 (15) is synergic with bubble 3(13), bubble 4 (14) , bubble 7 (17) and bubble 8(18).

[0159] Bubble 6 (16) is synergic with bubble 4(14)).

[0160] Bubble 7 (17) is synergic with bubble 5(15).

[0161] Bubble 8 (18) is synergic with bubble 5(15).

[0162] Bubble 9 (19) is synergic with bubble 2(12)

[0163] Using the synergic groups, the system can select, given a user query, all synergic bubbles within the tree for a reference bubble. For example, in the exemplary synergic groups of bubbles depicted on figure 10, if the query applies to reference bubble 15, then the system responds to the query with synergic bubbles 3, 4, 7, and 8.

[0164] It should be noted that in some embodiments, synergy is processed for plurality of trees rather than a single tree as described herein. Such implementations may apply in cases where there are multiple multi-media content enlarged by trees and one of these trees is that of the current multi-media content containing the reference bubble. In such case, the synergic groups of bubbles, as depicted on figure 10, shall include not only synergic groups from the current tree but rather synergic groups from the collection of all trees. An example could be the case where the current tree is a tree created to enlarge a medical video, and this medical video is a member of a library of medical videos for which trees were created. The collection of all such trees forms a domain of trees related to medicine, and the synergy is processed across one or more of the trees within the domain.

[0165] Figure 11 shows a flowchart diagram of the Learn and Deduct process, in accordance with some embodiments of the disclosed subject matter.

[0166] According to some embodiments the system records and analyzes the knowledge mining activities of end users during their knowledge mining sessions. The recording and analysis are for providing history of end users’ behavior, for identifying recommended multi-media content, for generation of various analytics of end users’ behavior, and for predicting future trends of knowledge consumption given past trends.

[0167] According to some embodiments each bubble and each meta-knowledge item includes a counter. The system increases the counter each time the end user visits the bubble and / or a metaknowledge item while performing knowledge mining and, in addition, the system may record the end user’s identifier, the location of the end user, the time stamp of the visit, and total time elapsed on each knowledge mining session. The system also records each operation of the end user on the dashboard. Such recording can be used later on by the system to re-construct knowledge mining sessions that have already occurred and provide the user with access to such information.

[0168] The Learn and Deduct process identifies bubbles that are recommended. Using the counter of visits of each bubble, Learn and Deduct sets the status of a bubble to recommended if the counter exceeds or equals to a pre-defined threshold and / or exceeds or equals to a predefined percent above the calculated mean of all counters of visits in all bubbles.

[0169] The Learn and Deduct process enables the content creator with the composition of analytics rules for performing analytics queries on the data associated with end users’ behavior and generate various analytics reports. Queries may include, for example, “what are the most recommended ten bubbles” or “provide me with a sorted list of bubbles visited more than 1000 times in the past three months” or “how many end users have visited bubbles about a cataract procedure” and so on.

[0170] The analytics rules for queries can be composed as free-text or using a query syntax such as {RECOMMENDED BUBBLES, TIME-PERIOD=<3, COUNTER>= I OOO. SORTED}, which defines, for example, a query for all recommended bubbles in the past three months with counters of at least 1000 visits, sorted by the visit counter top-down. An Al method such as NLP can be used to analyze a rule entered as free-text and construct the query syntax.

[0171] The Learn and Deduct process also enables the content creator to improve the bubbles’ tree by defining analytics rules related to trends in end users’ behavior and performing queries for such rules. Such trend analysis can identify among all bubbles, for example, bubbles that on average are rarely visited by end users. Such bubbles can then be removed by the content creator or enlarged to create more end user traction. An example of trend analytics rule may be of the syntax {BUBBLES, AVG(COUNTER)<10%} which generates as a response all bubbles that have a counter which is 10% or less than the average visit counter of all other bubbles.

[0172] Referring now to the drawing:

[0173] Blocks 1100, 1200 and 1300 describe the process that is associated with the content creator for defining and applying analytics rules. At block 1100 the content creator defines one or more rules for generating analytics. Such rules may identify, for example, cases in which a counter of one or more meta-knowledge items is greater than a certain threshold during a certain time-period. The syntax for such rule may be, for example, {BUBBLES, COUNTER>=1000, TIME-PERIOD<=6 } or in free-text “show me all the bubbles visited at least 1000 times during the past six months”. In case the analytics rule is entered as free text an Al method such as NLP can construct the syntax of the rule from the free text.

[0174] At block 1200 the content creator applies the rules on the meta-knowledge items of bubbles related to end users’ behavior. In some embodiments the system applies the rules automatically on a periodic basis to enable content creators generate the analytics results in advance.

[0175] As a result, at block 1300 the system generates a list of bubbles and / or meta-knowledge items that comply with each of the rules.

[0176] Blocks 1400, 1500, 1600, 1700 and 1800 illustrate scenarios for utilizing the Learn and Deduct process.

[0177] Blocks 1400 and 1500 illustrate scenarios for utilizing the Learn and Deduct process by the end user.

[0178] At block 1400 the user posts a query via the dashboard. A free-text example for such a query may be “show me recommended bubbles about heart failure”. An Al method such as NLP enables conversion of the free-text query to a syntax query. On one example, such syntax query may be {RECOMMENDED BUBBLES, kcyword=”hcart failure”}.

[0179] At block 1500 the system presents the results of the query. On one example the multimedia content of the recommended bubbles in the tree containing the string “heart failure” in any of their meta-knowledge items are presented.

[0180] Blocks 1600 and 1700 and 1800 illustrate scenarios for utilizing the Learn and Deduct process by the content creator.

[0181] At block 1600 the content creator posts a rule such as “what are the bubbles viewed less than 10 percent of the average of all other bubbles in the recent year”. An Al method such as NLP enables conversion of the free text rule to the syntax {BUBBLES, AVG(COUNTER)<10%, PERIOD= I 2 ).

[0182] At block 1700 the system applies the query according to the rule on the tree and identifies meta-knowledge items that comply with the query and presents results.

[0183] At block 1800 the content creator operates in accordance with the results. For example, the content creator may remove the least viewed bubbles from the tree. In some embodiments, the system operates in accordance with the results automatically. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and / or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0184] It should be noted that, in some alternative implementations, the functions noted in the block of a figure may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved.

Claims

CLAIMSWhat is claimed is:

1. A computer-implemented method for enlarging multi-media content, the method comprising: creating a root of a tree for the multi-media content; receiving a second multi-media content; the second multi-media content being retrieved by a computerised search; the computerised search utilizing one or more keywords related to a media content object; the media content object being identified by tagging of the multi-media content; generating a child node of the tree; the child node comprises the second multi-media content; and linking the child node to the root to thereby enlarging the existing multi-media content with the second multi-media content.

2. The method of claim 1, further comprising presenting the second multi-media content on a computing device of a user.

3. The method of claim 1, further comprising recursively repeating the generating the child node and the linking for generating one or more branches of the tree for enlarging each of the branches of the tree with enlarged multi-media content related to a parent of the branches.

4. The method of claim 1, further comprising increasing a value of a counter of the child node, the counter being for counting visits of one or more end users in the child node and presenting the child node only if the counter being above a threshold.

5. The method of claim 1, further comprising: receiving a similarity query from a user, the query comprises one or more reference meta-knowledge items; traversing the tree of the enlarged multi-media content or one or more trees including the tree of the enlarged multi-media content ; identifying similar multi-media content complying with the similarity query, and presenting the similar multi-media content to the user.

6. The method of claim 5, wherein the identifying of the similar multi-media content being if the reference meta-knowledge items of the similarity query being similar according to the similarity query to one or more second meta-knowledge items of a same type associated with the similar meta-knowledge items of the similarity query.

7. The method of claim 1, further comprising receiving a synergy query from a user for presenting synergic multi-media content, the query comprises a reference multi-media content; traversing the tree of the enlarged multi-media content or one or more trees including the tree of the enlarged multi-media content; identifying synergic multi-media content, and presenting the synergic multi-media content to the end user.

8. The method of claim 7, further comprising retrieving, from a data repository, a rule associated with synergy, the rule dictating a proximity level with related operands between an at least one meta-knowledge item of each multi-media content of a pair of nodes; identifying in a data repository a synergic group associated with the reference bubble; the synergic group comprises one or more synergic nodes; the each of the synergic nodes and the reference bubble comply with the rule; and presenting to the user the synergic nodes.

9. The method of claim 8, wherein the synergic group being generated by a process of identifying proximity level between at least one meta-knowledge item of each pair of nodes in the tree and applying the synergy rule on each pair of nodes; whereby synergic nodes are defined as those that comply fully with the rule.

10. The method of claim 1, further comprising applying a rule associated with past behavior of the end users with the tree pertaining to the number of visits of the multi-media content and its parameters, and the time of visits; extracting one or more multi-media content from the tree in accordance with the rule, the one or more multi-media content identifying trends in the end users’ behavior, and presenting the one or more multi-media content or the trends on a computing device.

11. A computer-implemented method, the method comprising: receiving a query from a user for presenting synergic multi-media content, the query comprises a reference multi-media content; retrieving, from a data repository, a rule associated with synergy, the rule dictating a proximity level between an at least one meta-knowledge item of two bubbles in a tree; identifying in the tree of the enlarged multi-media content or in one or more trees including the tree of the enlarged multi-media content a synergic group associated with the reference bubble; the synergic group comprises one or more synergic bubbles; each of the one or more synergic bubbles and the reference bubble comply with the rule; and presenting to the user the synergic bubbles.

12. The method of claim 11, wherein the meta knowledge item being multi-media content associated with the bubble or a parameter describing the multimedia content.

13. The method of claim 1, further comprising retrieving of a second multi-media content by a computerised search of one or more bubbles’ trees; the computerized search utilizes metaknowledge items defined in a query.

14. A system for creating a tree for enlarging multi-media content, the system comprising: one or more processors; and a memory communicatively coupled to the one or more processors,wherein the memory comprises instructions which, when executed by the one or more processors, cause the one or more processors to perform a method comprising: creating a root of a tree for the multi-media content; receiving a second multi-media content; the second multi-media content being retrieved by a computerised search; the computerised search utilizing one or more keywords related to a media content object; the media content object being identified by tagging of the multi-media content; generating a child node of the tree; the child node comprises the second multi-media content; and linking the child node to the root to thereby enlarging the existing multi-media content with the second multi-media content.

Images