WO2021071428A1

WO2021071428A1 - System and method for innovation, creativity, and learning as a service

Info

Publication number: WO2021071428A1
Application number: PCT/SG2020/050570
Authority: WO
Inventors: Chien Wei CHIA; Bhagwan Jethanand DASWANI
Original assignee: Ai Robotics Limited
Priority date: 2019-10-09
Filing date: 2020-10-08
Publication date: 2021-04-15
Also published as: TW202127364A

Abstract

A system and method for self-learning in innovation, creativity, and learning to facilitate users' task performances. The system is configured to better enable the construction of knowledge frameworks in relation to the tasks, determine a relevant goal; suggest a question based on the knowledge framework; and enable the user to input an answer and an insight. The system may be further configured to with a user interface that provides an emergent labelling tool and/or a knowledge visualization tool. The system is configured to provide a software as a service to enable creation of better knowledge frameworks, which in turn can be used to accelerate organizational wide self-learning for improved task performance.

Description

SYSTEM AND METHOD FOR INNOVATION, CREATIVITY, AND LEARNING AS A

SERVICE

TECHNICAL FIELD

[0001] The present disclosure generally relates to the field of artificial intelligence epistemological systems and methods, and more to a system and method for providing innovation, creativity, and learning as a service.

BACKGROUND

[0002] From an epistemological perspective, people (including individuals and teams of individuals) can potentially improve their performance of a task if they have a better knowledge framework. However, in reality, many people may need to acquire relevant knowledge frameworks before they start performing a task for the first time. Some others may have notions of relevant knowledge frameworks from their experiences but are constantly seeking and contextually constructing better ones to improve their task performances . Given the complexity of these nature of knowledge work, the contextual acquisitions and constructions of such improved knowledge frameworks exact a certain demand on learning, innovation and creativity from those involved.

SUMMARY

[0003] In one aspect, the present disclosure describes a system for self-learning in innovation and creativity to improve a user’s performance of a task based on information derived from one or more source objects. The system includes: a non-transitory computer-readable storage medium containing instructions; and a system server coupled to the storage medium. The system is configured to perform the instructions for a method including: providing a user interface, the user interface being configured to receive user input; and constructing a knowledge framework in relation to the task. The knowledge framework is defined by a plurality of historical first attribute labels, each of the plurality of historical first attribute labels being associated with at least one corresponding historical contextual element. The method further includes providing an auto-suggestion based on a contextual element and the knowledge framework, the auto-suggestion including one or more candidate attribute labels generated by the system in response to the contextual element being selected at the user interface, wherein each of the one or more candidate attribute labels is corresponds to one of the following: a goal relevant to the task, a question related to the task, an answer to the question, and an insight that is different from the answer.

[0004] Optionally, providing the auto-suggestion includes: in response to the contextual element being selected at the user interface, determine a first attribute label, the first attribute label being associated with the contextual element such that the first attribute label is defined relative to the knowledge framework, the contextual element being based on the one or more source objects; providing the one or more candidate attribute labels as the auto-suggestion; and in response to the auto-suggestion being received as user input, enabling the user to input one or more other attribute labels such that the contextual element is associated with a label. The label is defined by at least the first attribute label, the second attribute label, the third attribute label, and the fourth attribute label, wherein the first attribute label corresponds to a goal relevant to the task, the second attribute label corresponds to a question related to the task, the third attribute label corresponds to an answer to the question, and the fourth attribute label corresponds to an insight that is different from the answer. The system may be configured such that the auto-suggestion includes the candidate attribute labels corresponding to the second attribute label and the third attribute label. The system may be configured such that the auto suggestion includes the candidate attribute labels corresponding to the first attribute label, the second attribute label, and the third attribute label. The system may be configured such that the auto-suggestion includes the candidate attribute labels corresponding to the second attribute label, the third attribute label, and the fourth attribute label. The system may be configured such that the auto-suggestion includes the candidate attribute labels corresponding to the first attribute label, the second attribute label, the third attribute label, and the fourth attribute label. The system may be configured such that the auto-suggestion includes candidate attribute labels corresponding to any combination of one or more of the following: the first attribute label, the second attribute label, the third attribute label, and the fourth attribute label.

[0005] Optionally, the system further includes: providing a user interface having an emergent labelling tool, the emergent labelling tool being configured to receive user input in response to the contextual element being selected, wherein the user input includes an attribute label, and wherein the attribute label is one of the following: the first attribute label, the second attribute label, the third attribute label, the fourth attribute label and/or one or more customized labels.

[0006] Optionally, providing the auto-suggestion for the candidate second attribute label includes: identifying a historical first attribute label from the plurality of first historical attribute labels based on a degree of similarity between respective associated contextual elements; and using a historical second attribute label that is associated with the historical contextual element as a candidate second attribute label of the auto-suggestion. Optionally, the knowledge framework includes historical first attribute labels received as user input from at least a first user, and wherein the auto-suggestion is provided to a second user.

[0007] Optionally, the auto-suggestion for the candidate second attribute label is based on a historical second attribute label or a generative second attribute label. The historical second attribute label is associated with a historical contextual element having a degree of similarity with the contextual element, the historical contextual element being associated with a historical first attribute label that is part of the knowledge framework. The generative second attribute is generated by a method including: based on the contextual element as an input, obtaining an output from a natural language processing module; and iteratively comparing the output against a historical second attribute label using a generative adversarial network, wherein the historical second attribute label is selected based the first attribute label and the knowledge framework. Optionally, the knowledge framework includes historical first attribute labels received as user input from at least a first user, and wherein the auto-suggestion is provided to a second user.

[0008] Optionally, the system is further configured to generate a label based on a weighted ensemble of trained machine learning models, wherein a greater weightage is assigned to the model of the historical fourth attribute labels relative to respective weightages assigned to any one of the models of the historical first attribute labels, historical second attribute labels, and historical third attribute labels in the weighted collection of labels.

[0009] Optionally, the emergent labelling tool further includes: a manual interactive panel configured to receive manual user input; and an auto-suggestion interactive panel configured to provide the auto-suggestions.

[0010] Optionally, the user interface further comprises a knowledge visualization tool configured to provide one or more auto-suggestions, wherein the one or more auto-suggestions are provided in response to a new source object being provided by one or more social media channel subscriptions linked to the task.

[0011] According to another aspect, the present disclosure includes a method for deploying a software as a service according to any of the embodiments described above. BRIEF DESCRIPTION OF DRAWINGS

[0012] Fig. 1 is a schematic diagram of a system for innovation, creativity, and learning as a service.

[0013] Fig. 2 is a schematic diagram of a user interface having an emergent labelling tool, according to one embodiment of the system.

[0014] Fig. 3 illustrates a knowledge framework based on goal attribute labels.

[0015] Fig. 4 is a schematic block diagram of the system according to one embodiment.

[0016] Fig. 5 is a schematic process diagram of a machine learning module of the system.

[0017] Fig. 6 is a schematic diagram of the emergent labelling tool, according to another embodiment.

[0018] Fig. 7 is a schematic diagram of the emergent labelling tool, according to another embodiment.

[0019] Fig. 8 is a schematic diagram of the emergent labelling tool, according to another embodiment.

[0020] Fig. 9 is a schematic diagram of a knowledge visualization tool provided by the system.

[0021] Fig. 10 is a schematic flow diagram of a method according to embodiments of the present disclosure.

[0022] Fig. 11 is a schematic flow diagram of a method according to other embodiments. DETAIFED DESCRIPTION

[0023] It will be readily understood that the components of the embodiments, as generally described and illustrated in the figures herein, may be arranged and designed in a wide variety of different configurations in addition to the described example embodiments. Thus, the following description of the example embodiments, as represented in conjunction with the figures, is not intended to limit the scope of the embodiments, as claimed, but is merely representative of example embodiments. [0024] Reference throughout this disclosure to “one embodiment”, “another embodiment” or “an embodiment” (or the like) means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. Thus, the appearance of the phrases “in one embodiment” or “in an embodiment” or the like in various places throughout this specification are not necessarily all referring to the same embodiment.

[0025] Furthermore, the described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a thorough understanding of embodiments. One skilled in the relevant art will recognize, however, that the various embodiments may be practiced without one or more of the specific details, or with other methods, components, materials, etc. In other instances, some or all known structures, materials, or operations may not be shown or described in detail for the sake of clarity.

[0026] [0010] Embodiments of a system and method for providing self-learning in innovation, creativity, and learning as a service will be described with reference to a work environment, solely to aid understanding. It can be understood that embodiments of the present disclosure can also be applied to various different application scenarios, such as learning, work, personal, and/or social situations. In one exemplary scenario, a first user (for example, a designer) is assigned to create a design for the product packaging of a new product. The new product is a new innovation such that the first user has no prior knowledge or experience related packaging requirements for the product. A second user in the company is assigned to execute a digital marketing campaign for the new product. The second user is new to the process of digital marketing. The second user also has no prior knowledge or experience in either the subject matter of digital marketing or the new product. Such a scenario is an example of a sterile case where the learners lack a relevant epistemic system. It can be appreciated that a lot of learning, innovation, and creativity is required from individuals and teams in order for the company to kickstart a digital marketing campaign and launch the new product in the market. Therefore, the conventional way of acquiring knowledge relevant to a task urgently needs to improve.

[0027] According to one aspect, a system 100 according to an embodiment of the present disclosure provides a tool 200 for creating better knowledge frameworks 300, which in turn can be used to accelerate learning and improve task performance. The system 100 is configured so that it can provide a service for promoting self-learning and/or improving task performance. [0028] According to an embodiment, as shown in Fig. l, asystem 100 for creativity, innovation, and learning as a service may be implemented by a system server 110 having at least one processor configured to execute instructions stored on computer-readable storage medium 112. The system 100 is configured to provide a service that can be accessed over a network 120 by a user through a user device 130 (such as a mobile phone, a tablet, a laptop, or a computing or smart device). The system may be configured to store data in a database 140, such as in a client database. Third-party sources 150 of information may be referred to in the course of a user’s learning journey, for example, a website. For the avoidance of doubt, reference in this document to a user’s access to such third-party sources refers to authorized access to such sources.

[0029] Fig. 2 shows an example of a user interface 200 provided by the system 100. The user interface 200 may be in the form of a software application configured for use on a computing device. The user interface 200 may be in the form of a plug-in, an add-on, utility program. The user interface may be provided as part of an on-line platform, a cloud-based application, a downloadable program, etc. The user interface 200 may be configured for use with a browser, an e-book reader, or other applications, such as social media apps, learning platforms, news apps, etc. These possible forms are given merely as examples.

[0030] The user interface 200 includes an emergent labelling tool 210. In this embodiment, the emergent labelling tool 210 provides a text box 220 for user input of labels 300. As the first user reads about product packaging online (in this example, using a web browser). The emergent labelling tool is configured to enable the first user to select (for example, by highlighting) one or more contextual elements 230 (in this example, the phrase “you are curating ideas for the ideal customer and not for yourself’) from a representation of a source object 240 (in this example, a web article). The emergent labelling tool 210 is shown here as a pop-up menu, but it can be in the form of a plug-in, an add-on, or a utility program, etc.

[0031] The source object may correspond to an electronic book, a PDF document, a video, an audio recording, a multimedia file, a website, a blog, a vlog, an app, a social media posting, or any electronic entity. The source object may be accessed through an electronic means of presenting, storing, or conveying information in a form that can be read, heard, or otherwise sensed by the user. The user interface 200 is operable with other applications so that the source object or the information therein may be accessed by the user without transgressing rights permitted by a copyright owner of the electronic file. For instance, the user interface 200 is configured so that the user may access the electronic file through the usual permitted venues such as, for example, an internet browser, an e-book reader, an electronic library membership account, a social media application, etc. The source object may thus be one residing locally in the user’s computing device 130 or one residing in another part of a network 120. The network 120 may be accessible over a wireless, cabled or another form of connection. Examples of the network 120 include a local area network, a wide area network, an extranet, an intranet, etc. The source need not be an electronic file. For example, the source may be an identifiable printed publication. In some embodiments, the user interface 200 may provide a referencing tool so that the user may identify a non-electronic source, e.g., by referencing an international standard book number (ISBN) or a journal article, etc.

[0032] Responsive to the selection of a contextual element 230, the system 100 is configured to associate the contextual element with at least one label 300/310. The system is configured to receive, suggest, and/or generate the at least one label 300/310. The system is similarly able to receive input from the second user, for example, the second user can use the emergent labelling tool 210 to create contextual elements relating to digital marketing campaigns, in which each contextual element 230 is associated with at least one corresponding label 300/310.

[0033] Each label 300 may be defined in terms of one or more attribute labels 310. In this document, the term “label” (or “labels”) can refer to attribute labels, or the term can refer to labels defined by one or more attribute labels. The system 100 is configured to store the labels 300/310 in a database 140 as a collection of labels, in which each label 300 includes one or more attribute labels 310. The system 100 is configured such that each attribute label 310 can be associated with corresponding contextual elements 230.

[0034] In this example, the label 300 can be defined in terms of one or more of the following attribute labels 310: a goal attribute label 312 (also referred to as a first attribute label), a question attribute label 314 (also referred to as a second attribute label), an answer attribute label 316 (also referred to as a third attribute label), and an insight attribute label 318 (also referred to as a fourth attribute label). For example, in relation to the selected contextual element 230, the first user may use the emergent labelling tool 210 to input a label 300, in which the label 300 is defined in terms of the following attribute labels 310: label = (goal, question, answer, insight) goal = preliminary design question = who are my ideal customers? answer = Australian international schools insight = koala bears

[0035] The emergent labelling tool 210 may be further configured so that the user can label one or more contextual elements 230 with one or more labels 300, in which the labels include an answer attribute label 316. Each answer attribute label 316 points to or corresponds to an answer. The answer attribute label 316 corresponds to specific question attribute labels 314. The system 100 is configured to enable the user (the learner) to devise an answer attribute label 316, and to relate the answer label to a selected contextual element 230. Each of the answer attribute label 316 and the corresponding question attribute label 314 are related to at least one contextual element 230.

[0036] The emergent labelling tool 210 may be further configured so that the user can label one or more contextual elements 230 with one or more insight attribute labels 318. Each insight attribute label points 318 to or corresponds to an insight. An insight may be described as a cognitive event when the user experiences an “Aha!” or “Eureka!” moment. The user who inputs or selects an insight attribute label 318 is the learner, that is, the person who is undergoing the learning journey. The system 100 is configured to enable the user (the learner) to devise an insight attribute label 318, and to input or select the insight attribute label 318 to a selected contextual element 230. In one aspect, the insight attribute label 318 can be used to capture a creative or innovative thought. In one aspect, the insight attribute label 318 “koala- bears” is associated with a pair of labels, in which the pair of labels include a question attribute label 314 “who are my ideal customers?” and an answer attribute label 316 “Australian international school”. In another aspect, the insight attribute label 318 is associated with at least one first attribute label and at least one other label. In this example, the goal attribute label 312 is configured as the first attribute label 410. The question attribute label is the at least one other label 310.

[0037] As shown in Fig. 3, the system 100 is configured such that a framework 400 of nodes 410 may be organized based on one of the various types of attribute labels 310. In this document, such labels are also referred to as first attribute labels 312. The first attribute labels 312 are configured such that they can form nodes 410 in a hierarchical framework 400 based on relationships between respective framework labels 410. In this example, the goal attribute labels 312 are configured as first attribute labels 410. The framework 400 can be defined as a network of nodes 410, in which each node corresponds to a first attribute label 312. Each goal attribute label 312 may be configured to carry information on its relative position in the framework 400. The framework 400 may be organized progressively or iteratively. Relationships may be formed between the first attribute labels/nodes 312/410, based on their degree of relatedness. The system is configured to enable construction and/or modification of the framework 400 in response to labels 300 being added into a collection of labels, and as a result of relationships being formed between respective framework labels/nodes 312/410. The relationships between respective framework labels/nodes 312/410 can be used to represent a knowledge framework 400 for the users. In this example, the knowledge framework 400 based on labels input by the first user can be used to help the first user accelerate his learning for the purpose of designing a packaging for the new product.

[0038] At the same time, when the second user searches for tips on planning a digital marketing campaign, he may identify different contextual elements 230 based on different source objects 240 and associate them with different labels 300. For example, as the second user reads about the use of consumer behaviour analysis in marketing, he may have questions about the relevance of consumer behaviour analysis to his specified task, that is, launching a digital marketing campaign for the new product. The system 100 enables the second user to input these questions in the form of labels 300 having at least a question attribute label 314 associated with contextual elements 230, in which the contextual elements are based on the reading materials of the second user. In this case, the second user may start with several labels, each having only one attribute label 310, such as: question = when to launch campaign? question = will social media be relevant for this product? question = who is the target market?

[0039] Each of the contextual elements selected by the second user will be associated with a label that includes one or more blank attribute labels, for example: label = (blank, when to launch campaign, blank, blank)

[0040] The system 100 is configured such that users can continue to embellish existing labels 300. For example, the second user may retrieve the question he previously input to the system in relation to a contextual element. The second user can further embellish the label 300 so that the label 300 is now defined by more attribute labels 310, such as: question = when to launch campaign? answer = holiday season

[0041] As shown in this example, the questions (input for the question attribute label 314) are posed by the first user and the second users themselves. That is to say, the question input is provided by users when they are in the role of learners (the users who are undergoing a learning journey to acquire knowledge). Every question attribute label 314 that is input, suggested, or generated is one that corresponds to or is associated with corresponding contextual elements 230.

[0042] As the second user continues to leam about the subj ect matter over time, he can continue to add new labels to his collection of labels. The system 100 is configured to store a second collection of labels originating from the second user, such that the second user may have his own personal knowledge framework. This knowledge framework 400 can be used to represent a current state of personal knowledge associated with the second user. The system 100 may also be configured so that both the first user and the second user contribute to a shared collection of labels. The resulting knowledge framework 400 can be used to represent a current state of collective knowledge of a team of users.

[0043] According to one aspect, the system 100 is configurable to provide a service to different clients. A client (such as a company) may have multiple employees using the system 100 as learners. The client may appoint one of its employees as a facilitator, or an administrator. The facilitator or administrator may also be a learner. An administrator refers to a person or persons who manage administrative aspects of the learning, e.g., track attendance, provide access to resources, etc. A facilitator refers to a person or persons who design a syllabus and/or deliver a teaching content. The user interface 200 is configured so that the role of a facilitator/administrator will not be foregrounded unless necessary. For example, the provision of a pre-determined syllabus by a facilitator to the learner is optional. The system 100 is therefore advantageously suitable for real life or continuous learning where the learner may be required to start from a state where there is no prior knowledge and no prior syllabus or framework to help the learner leam systematically, towards achieving a goal. The learner may also be required to “leam on the job” without the benefit of guidance from a human mentor or a manual. In one embodiment, the user (including the learner) may customize or set-up other customized attribute labels 310 and/or framework labels 410 such as “important”, “take note”, etc. In other words, while organizations in the education and training industry are not precluded as users of the services provided by the system 100, actually the system 100 is suitable for a greater variety of broader application scenarios.

[0044] According to another aspect, the system 100 enables the learner to create his own dynamic “textbook” or library of information sources 150. In the example of the first user, the first user may crawl the internet for images of product packaging for similar products. The first user may visit a website and identifies a relevant standard as a first contextual element. The emergent labelling tool 210 enables the first user to identify a contextual element 230 by identifying its location, e.g., URL (uniform resource locator), and/or other identifiers. The emergent labelling tool 210 may enable the first user to identify a contextual element using a snapshot tool. The emergent labelling tool 210 may also be further configured to enable the user to copy a portion of the source as a contextual element 230, in which the maximum size of the portion that can be copied is controllable or predetermined according to what is permissible or to what is permitted by the copyright owners subscribing to the system.

[0045] According to yet another aspect, the first user may create a new question attribute label 314 and apply the question attribute label 314 to the identified/selected contextual element 230. The system 100 stores the question attribute label 314 in a database 140 as part of a collection of labels. If, after some use of the system 100, the first user has developed a database 140 made up of a plurality of question attribute labels 314, the first user may subsequently select a question attribute label 314 from the collection of labels, and relate the selected question attribute label 314 to another contextual element 230. According to one aspect, the user is “compelled” to think as a learner, i.e., as a seeker of knowledge, as the user is required to think in terms of questions. By being the one formulating questions (as opposed to focusing on answers or pieces of information), the learner is tuned to recognize his/her own knowledge gap(s). Additionally, the system 100 recognizes that there is more learning that can be derived from a question than from an answer. The system 100 is configured to extract from a question a piece of the puzzle which when assembled with other pieces of the puzzle produces at least a partial knowledge framework 400.

[0046] The various aspects described above differ from a traditional learning system in which a facilitator (who has prior knowledge in a subject matter) designs a syllabus (pieces of knowledge to be taught, sequence of teaching the pieces of knowledge, assessment of the learner’s absorption or retention of the pieces of knowledge taught), the system 100 according to the present embodiments provide for learning more consistent with the spirit of self-directed, self-paced, and self-actualized learning. Rather than inundating a learner with information selected by a facilitator, the system 100 is configured for the learner to drive his/her own learning. In particular, the system 100 is configured to enable dynamic evolution of a knowledge framework 400. Such a knowledge framework 400 will be configured according to at least one goal set or task of the learner. Such a knowledge framework 400 is designed to help the learner perform a task or improve in performing a task. According to this embodiment, the system 100 is configured such that the knowledge framework 400 is based on framework labels 410 associated with question attribute labels 314.

[0047] The system 100 is configured to build a knowledge framework 400 based on other labels 300/310, such as, answer attribute labels, insight attribute labels, goal attribute labels, including labels defined by any one or more of these attribute labels. A knowledge framework 400 may also be based on customized labels 300/310, that is, types of labels customized by the user. The knowledge framework 400 may be based on other attribute labels 310 besides those mentioned above, for example, vision-related attribute labels, purpose-related attribute labels, etc. A label 300 may accordingly be defined by one or more attribute labels 310. A label 300 may be made up any number of attribute labels 310.

[0048] According to another aspect of the present disclosure, the system and method is implemented with artificial intelligence/machine learning algorithms 520. Fig. 4 schematically illustrates another aspect of the system 100. The system 100 is further configured with machine learning modules 500. The system is configured with one or more databases 140 storing contextual elements 230 associated with respective labels 300/310 in one or more collections of labels, and related knowledge frameworks 400. The machine learning algorithms 520 are configured to draw on the data stored in the databases 140. One example is a natural language processing (NLP) engine that is configured to extract data from the identified/selected contextual elements 230 and the applied labels 300/310. As another example, the first user may also use the user interface (such as one shown in Fig. 2) to label what he perceives to be an important part of text/video/image/audio file (source) with two labels “efficient” or effective”. The meaning of the contextual elements and of the labels can be extracted by the NLP engine and fed to machine learning modules 500 for supervised or even unsupervised learning as shown in Fig. 5.

[0049] As shown in Fig 5, a machine learning module 500 of the system 100 may be configured to include one or more sub-modules under supervised or unsupervised learning. The machine learning module 500 may be configured to include one or more sub-modules, in which the sub- modules may be configured to implement one or more of the following in combination: artificial neural network model, deep learning algorithm, dimensionality reduction method, feature -extraction method, embedding-generation method, ensemble method, instance-based method, Bayesian method, clustering, and other suitable methods. The machine learning module may include a sub-module configured to implement one or more artificial neural network models. Examples of the artificial neural network models include, but are not limited to: Backpropagation method, Hopfield network method, Self-organizing Map method, and a Learning Vector Quantization method, etc. The machine learning module may include a sub- module configured to implement one or more deep learning algorithms. Examples of the deep learning algorithms include, but are not limited to: Transformers, LSTMs, Auto-encoders and Decoders, Generative Adversarial Networks, Deep Belief Network Method, Convolutional Neural Network, Recurrent Neural Network, Stacked Auto-Encoder, etc. The machine learning module may include a sub-module configured to implement one or more dimensionality reduction methods. Examples of the dimensionality reduction methods include, but are not limited to: principal component analysis, partial least squares regression, Sammon mapping, multidimensional scaling, projection pursuit, etc. The machine learning module may include a sub-module configured to implement one or more feature-extraction methods. The machine learning module may include a sub-module configured to implement one or more embedding- generation methods. Examples of the feature-extraction methods and/or embedding -generation methods include, but are not limited to: Continuous Bag of Words (CBOW), Skip-gram, etc. The machine learning module may include a sub-module configured to implement one or more ensemble methods. Examples ofthe ensemble methods include, but are not limited to: boosting, stacked generalization, gradient boosting machine method, random forest method, etc. The machine learning module may include a sub-module configured to implement one or more instance-based methods. Examples of the instance-based methods include, but are not limited to: k-nearest neighbour, self-organizing map, etc. The machine learning module may include a sub-module configured to implement one or more Bayesian methods. Examples of the Bayesian methods include, but are not limited to: naive Bayes, Bayesian belief network, etc. The machine learning module may include a sub-module configured to implement one or more clustering methods . Examples of the clustering methods include, but are not limited to: k-means clustering, expectation maximization, etc. [0050] The machine learning module in Fig. 5, can be configured to leam from the contextual elements and labels to embellish or enrich the knowledge frameworks. Specifically, the system 100 may be configured to suggest relevant labels 300/310 pertaining to the various aspects of the framework 400 being acquired or built. The system 100 can be configured to suggest various types of labels (auto-suggestion). Examples of two types of auto-suggestion (historical and generative) will be described below.

[0051] Referring to Fig 6, by this time, based on input from the first user, the database 140 has accumulated a collection of labels and a knowledge framework 400 has been generated. The system 100 is configured to automatically suggest candidate labels and/or candidate attribute labels. The candidate labels and/or candidate attribute labels may be presented to the user through the emergent labelling tool 210 or the user interface 200. The auto-suggestion may be presented in response a contextual element being selected. In this document, the terms “candidate label” and “candidate attribute label” may be used interchangeably as understood from the context, and refers to a label/attribute label which is auto-suggested by the system to the user. If the user accepts the auto-suggestion, the candidate label or candidate attribute label is received as user input and used as the corresponding label/attribute label.

[0052] According to one embodiment, the emergent labelling tool 600 is configured to suggest candidate labels 610 (or one or more candidate attribute labels 610). The candidate labels 610 suggested by the system 100 can include historical labels 300/310 retrieved from the database 140. The candidate labels 610 may be selected from historical labels 300/310 used in the same epistemic project, or the candidate labels 610 may be historical labels 300/310 from another existing or older project. The system 100 may be configured to select historical labels 300/310 based on a degree of similarity between the contextual element 630 selected by the second user and a historical contextual element 230 (Fig. 3). The historical contextual element 230 can include a contextual element which first user used in associating with at least one label. The historical contextual element 230 can also refer to contextual elements stored in the database. The historical contextual element 230 can also refer to contextual elements that are linked to a knowledge framework 400. In this example, the second user selects a contextual element containing “digital marketing begins with profiling your perfect client”. In response, the system 100 determines that this contextual element 630 has a certain degree of similarity with the historical contextual element “... curating design ideas ... for that ideal customer” 230 (Fig. 3). The system 100 presents the following as candidate labels 610 based on historical labels 300/310 associated with the historical contextual element 230: goal = preliminary design question = who are my ideal customers? answer = Australian international school insight = koala bears

[0053] In this manner, the second user can benefit from the first user’s learning. The emergent labelling tool configured with auto-suggestions 600 may be configured to allow the second user to confirm the use of the candidate labels 610. The emergent labelling tool 600 is also configured to allow the second user to modify or replace the candidate labels 610. In relation to the similar contextual elements, the second user may have different labels that are more relevant to the nature of his role or task. The system 100 is configured to capture these variations such that the resulting knowledge framework 400 is further enriched.

[0054] Referring to Fig. 7, according to another embodiment, in response to the second user selecting the contextual element 630 “digital marketing begins with profiling your perfect client” and providing a goal attribute label “focus”, the system 100 is configured to inform the second user that this may be related to an existing node 410 of a knowledge framework 400. The knowledge framework 400 may be one that is shared by the first user and the second user, because they are in the same team or working on the same or related tasks. If the second user selects a similar contextual element, the system 100 may suggest historical goal attribute labels and question attribute labels. The system 100 may be configured to provide such auto suggestions based on a machine learning module 500 as described above. Based on the knowledge framework 400, the system 100 may suggest to the second user that the historical goal attribute label “preliminary. design” may be related to the goal attribute label “focus”. The system 100 is configured to allow the second user to change the goal attribute label from “focus” to “preliminary. design.focus” 612. This attribute label 612 is also configured as a framework label 410, and the new label can be added to the relevant framework 400. The system 100 in effect helps the second user quickly understand how newly acquired knowledge can be related to earlier acquired knowledge, as well as enable the second user to quickly move on to acquire other knowledge, especially knowledge that is closely related to this label 612 in the knowledge framework 400. [0055] In response to a contextual element being highlighted, the system 100 can also be configured to suggest labels that are not found in the database. Fig. 8 shows another embodiment of the emergent labelling tool 800 which provides a first interactive panel 810 for manual user input of labels, a second interactive panel 820 for auto-suggestions based on historical labels, and a third interactive panel 830 for auto-suggestions based on a generative engine. In response to a contextual element being selected, a user may use a text box 815 to input labels 300/310. The first interactive panel 810 is configured to enable the user to input different types of labels 300/310, for example, goal attribute labels 811, question attribute labels 812, answer attribute labels 813, insight attribute labels 814, and/or a combination thereof. In some embodiments, the emergent labelling tool 800 includes a manual input interactive panel 810 for manual user input of labels, and an auto-suggestion interactive panel 820/830 for providing auto-suggestions. Examples of different methods of providing auto suggestions are described below.

[0056] The second interactive panel 820 is configured to present auto-suggestions based on the selected contextual elements. The auto-suggestions may include goal attribute labels 821, question attribute labels 822, answer attribute labels 823, insight attribute labels 824, and/or a combination thereof. The emergent labelling tool 800 is configured to enable the user to adopt the auto-suggestions, reject the suggestions or modify the auto-suggestions for each attribute label 821/822/823/824. The emergent labelling tool 800is also configured to enable the user to manually input labels if the system does not provide auto-suggestions for any or all of the attribute labels. When the user accepts the auto-suggestions, the accepted auto-suggestions are deemed user input to the system. When the user modifies the auto-suggestions, the modified auto-suggestions are deemed user input to the system.

[0057] The third interactive panel 830 is configured to present auto-suggestions generated by a generative engine provided by the machine learning modules 500. In this disclosure, a generative engine could comprise a label generator based on a weighted ensemble of trained machine models arising from applying machine learning on the aforementioned historical attributes; or more differently with Generative Adversarial Networks. A generative engine using Generative Adversarial Networks will be described here as an example. In some embodiments, in response to the second user selecting a contextual element such as “digital marketing begins with profiling your perfect customer”, the system generates a candidate generative question and presents it to the user. An NLP (natural language processing/processor) module within machine learning modules 500 may be configured such that, based on the contextual element as an input, the NLP module outputs a possible question. Based on the relative position of the goal attribute label in a related knowledge framework, the generated question may be compared against a historical question, in which the historical question is selected based on its relevance according to the knowledge framework. If the possible question is determined by the system to be a “fake” question, this result will feedback to the generative engine. After iteratively performing this comparison and feedback, the system will present a candidate question that does not produce a “fake” result. In this example, the system provides the candidate question (candidate second attribute label) as an auto-suggestion. That is, the generative second attribute label “who are your loyal customers?” is presented to the second user as the candidate second attribute label. When the system is configured so, the users will be provided with more “out of the box” suggestions for labels. This allows new ideas to occur in the acquisition and building of new knowledge frameworks. In this example, the generated question pushes the user to adopt a slightly different perspective and to come up with a different answer from the suggestion based on historical labels. The different answer “Sydney International Schools” when considered together with the related goal and question, helps the user to make the cognitive connection to a new insight “Wallaby”.

[0058] The above is just one example of auto-suggestions for one of the attribute labels. In some other embodiments, one or more of the labels/attribute labels (for example, the goal attribute label, the question attribute label, the answer attribute label, the insight attribute label, and any customised labels) can be auto-suggested by the system, based on a selected contextual element and/or one of the other labels/attribute labels. Thus, when a user accepts a suggestion provided by the system (whether it is historical or generative), a service of the innovative type is said to be rendered because the process adds on to the knowledge framework 400. When a user accepts any other suggestions of the historical or generative type, it is said that a service of the creative type was rendered because it potentially opens up new perspectives for the users who are performing the task but not directly contributing to the embellishment of a knowledge framework per se.

[0059] According to another embodiment, in the example of the second user who has been tasked to execute a digital marketing campaign, the second user may use the system 100 to acquire knowledge about digital marketing. In the course of reading about the subject matter of digital marketing, the user is able to label the types of problems likely to be encountered in digital marketing. Overtime, the system 100 can leam what is important and not important, or what is more relevant and what is less relevant, in the context of the subject matter. Thus, over time, and with a growing library of sources (references to sources) 140/150 represented by contextual elements associated with labels, the system 100 can go further to play the role of a facilitator to a user. For example, on and off, the system 100 may pose a question to the user, e.g., “What is clearer in your mind about digital marketing?” or more generally, “What were the most important points?”, “Have your questions been answered?”, “Are you able now to put them into project phases?”, or “What are the issues you think you will face?”, etc.

[0060] To further aid the organization of learning and to better embellish the various knowledge frameworks being constructed, the system 100 can be configured to execute (at the backend) a process of piecing together what has been labelled as “important” for example, or any other meaningful labels such as “efficient” or “effective”. These labels can also be weighted. In the aforementioned example, system 100 may be configured to take what is has been labelled “important” by the first user (e.g., use of sustainable packaging) to enrich or supplement the learning journey of the second user. The second user can thus benefit from the knowledge framework 400 built by the first user, and more important, not omit learning about an important feature for marketing the product.

[0061] According to another aspect, the system is configured to present its “findings” to the user, for example, through a graphical user interface format. Such “findings” could be configured to include areas for future learners to pay attention to from previously constructed knowledge frameworks. One schematic illustration of a knowledge visualization tool 900 is shown in Fig. 9. The “important” or “relevant” information may be presented to the user in a workbench environment configured for the user to re-arrange (select) 910 milestones, tasks, project phases, issues, problems, etc. 920, relative to a timeline 930 or a series of milestones 940. In this manner, the system 100 can better help the user to move from knowledge framework acquisition to application through the incorporation of the “important” information into timelines. The second user, for example, may use the user interface 200 provided by the system 100 to lay out the various steps of a digital marketing campaign project. The system 100 can be configured to present to the user links between questions and answers, and to link these to respective relevant steps/phases/milestones of a project. The system is thus a “live” environment. In one embodiment, for example, a task is linked to one or more social media channel subscriptions. The system 100 can be configured to present auto-suggestions to the user whenever new articles are being pulled from such a subscription service. In other words, the user interface is configured such that the knowledge visualization tool can provide one or more auto-suggestions in response to a new source object being provided by one or more social media channel subscriptions linked to the task.

[0062] If a third user (e.g., in another part of the world) is also interested in learning about digital marketing campaigns, the system 100 may accelerate the learning of the third user. Even without making known to the third user the question labels devised by the second user, the system 100 through its architecture is configured to help automate the identification of important contextual elements 230, as well as the suggestion of questions 314 and answers 316 in aid of learning. For example, over time, the system 100 may collect enough data to predict the kind of questions 314 that are more likely to lead a user to arrive at insights. The system can thus “nudge” subsequent users towards arriving at insights 318 (the “Aha! ” moment) earlier or more frequently. Over time, the system 100 is able to acquire a collection of such “high value” questions 314 for use in facilitating further creativity and learning.

[0063] The system 100 may be configured to provide analytics of the emergent labelling process. The system 100 may be configured to determine the number of respective types of labels 300/310. The system 100 may be configured to discover a chronological order of the labels 300/310 to aid visualization of the learning journey. For instance, in the digital marketing example, the system 100 can identify the length of time it took for a user to acquire an insight 318. As an example, the second user took 12 days to arrive at the insight 318 that the key in digital marketing is conversion. The system 100 can track the number of days it takes others to arrive at a similar insight. Over time, with the aid of the system 100, the subsequent learners can take significantly fewer days to arrive at a similar insight 318, owing to accelerated learning provided by the system 100. Overtime as more users use this system 100, the idea is to shorten the time from first learning to insights 318.

[0064] The system 100 is configured to track a learning outcome. That is to say, when given a task, the system 100 can track the amount of iterations the user takes to develop an appropriate knowledge framework 400. In the example here, the learning outcome is a new knowledge framework 400 and the system 100 is able to measure how long it takes for such frameworks to be constructed in organizations. In the example, this may take the form of providing the second user with the user interface 200 to plan a marketing campaign in terms of timelines corresponding to phases and sub-phases of the projects. The system 100 may also be used to identify the issues and problems in digital marketing likely to be encountered, in which the likely issues are based on the what is evolving dynamically as the second user leams. When the second user is ready, the second user can go to the next step which is to execute the campaign. The system 100 can continue to be used by the second user or another user to monitor the digital marketing campaign. The digital marketing campaign itself can serve as a source 240 of additional contextual elements 320 for further labelling, so that the knowledge framework 400 can be refined, improved, or otherwise updated. The emergent labelling tool 210 can thus continue to be used.

[0065] To aid understanding, the following describes one example of how the system 100 may have a broader application such that it may apply to a team of users. The system 100 may additionally be configured so that multiple users may participate in one learning journey. Input from respective members of the team may be color coded or otherwise identifiable/associated with respective members of the team. Team members can inspect the labels of other team members in the same team, and the user interface 200 may be configured to enable discussion without respect to the sources 240, contextual elements 230 and/or labels 300/310. The user interface 200 may also be configured so that a user can answer questions 314 posed by other members of the team. The emergent labelling tool 210 may be further configured to allow the user to link text/video/image/speech 240 to those questions 314.

[0066] The foregoing describes various embodiments of the system for self-learning in innovation and creativity towards improving a user’s performance of a task based on information derived from one or more source objects have been described above. The system includes a non-transitory computer-readable storage medium containing instructions and a system server coupled to the storage medium and to a user interface. The system is configured to perform instructions for a method (1000) as illustrated in Fig. 10. The method includes providing ( 1010) a user interface, the user interface being configured to receive user input. The method includes constructing (1020) a knowledge framework in relation to the task, the knowledge framework being defined by a plurality of historical first attribute labels, each of the plurality of historical first attribute labels being associated with at least one corresponding historical contextual element. The method further includes providing (1030) an auto suggestion based on a contextual element and the knowledge framework, the auto-suggestion including one or more candidate attribute labels generated by the system in response to the contextual element being selected at the user interface, wherein each of the one or more candidate attribute labels is corresponds to one of the following: a goal relevant to the task, a question related to the task, an answer to the question, and an insight that is different from the answer.

[0067] The system may be configured to perform the instructions for a method (1100) as illustrated in Fig. 11. The method includes constructing (1020) a knowledge framework in relation to the task, the knowledge framework being defined by a plurality of historical first attribute labels, each of the plurality of historical first attribute labels being associated with at least one corresponding historical contextual element. The method includes, in response to a contextual element being selected at the user interface, determining ( 1130) a first attribute label, the first attribute label being associated with the contextual element such that the first attribute label is defined relative to the knowledge framework. The contextual element is based on the one or more source objects. The method includes providing (1140) one or more candidate attribute labels as the auto-suggestion based on the contextual element and the knowledge framework. And the method includes, in response to the auto-suggestion being received as user input, enabling (1150) the user to input one or more other attribute labels such that the contextual element is associated with a label, the label being defined by at least the first attribute label, the second attribute label, the third attribute label, and the fourth attribute label, wherein the first attribute label corresponds to a goal relevant to the task, the second attribute label corresponds to a question related to the task, the third attribute label corresponds to an answer to the question, and the fourth attribute label corresponds to an insight that is different from the answer. The system may therefore be deployed in the form of a software as a service.

[0068] Knowledge frameworks as described in this disclosure are thus very specific and contextual, can take many forms and be continuously improved upon. In this disclosure, innovation, creativity and learning as a service are conceived as a system and method to thrust humans to more creatively and innovatively improve these knowledge frameworks so that all aspects of organizational performances can go to the next level and so on. In doing so, learning or the development of human potential whether as individual or teams is said to have taken place. From an epistemological perspective, this system and method enables the user to more systematically and rapidly acquire knowledge frameworks relevant to any specific task. Further, the form of the acquired knowledge frameworks will undergo evolution and transformation multiple times, changing from more abstract to more concrete forms, until the learner is satisfied that the resulting knowledge framework is contextually relevant. The emergent labelling tool enables better knowledge frameworks to be captured to facilitate future learning journeys. From an epistemological perspective, the data thus obtained presents a historical trail of how learning took place and could be referenced by future learners.

[0069] From another epistemological perspective, it is the insights that drive innovation and creativity. In particular, the above embodiments enable different users to benefit from relevant learning by other users, in which the relevant learning is automatically curated by the system. This is equivalent to a scenario where different users “brainstorm” together towards new innovations that could require a multitude of new insights to be generated, but has the advantage of enabling different users to participate online at different times from different locations.

[0070] As used herein, the singular “a” and “an” may be construed as including the plural “one or more” unless clearly indicated otherwise.

[0071] This disclosure has been presented for purposes of illustration and description but is not intended to be exhaustive or limiting. Many modifications and variations will be apparent to those of ordinary skill in the art. The example embodiments have been chosen and described in order to explain principles and practical application, and to enable others of ordinary skill in the art to understand the disclosure for various embodiments with various modifications as are suited to the particular use contemplated.

[0072] Thus, although illustrative example embodiments have been described herein with reference to the accompanying figures, it is to be understood that this description is not limiting and that various other changes and modifications may be effected therein by one skilled in the art without departing from the scope of the disclosure.

[0073] This application is related to Singapore patent application no. 10201909407W filed on 9 October 2019, which is incorporated by reference in entirety.

Claims

1. A system for self-learning in innovation and creativity to improve a user’ s performance of a task based on information derived from one or more source objects, the system comprising: a non-transitory computer-readable storage medium containing instructions; and a system server coupled to the storage medium, the system being configured to perform the instructions for a method comprising: providing a user interface, the user interface being configured to receive user input; constructing a knowledge framework in relation to the task, the knowledge framework being defined by a plurality of historical first attribute labels, each of the plurality of historical first attribute labels being associated with at least one corresponding historical contextual element; and providing an auto-suggestion based on a contextual element and the knowledge framework, the auto-suggestion including one or more candidate attribute labels generated by the system in response to the contextual element being selected at the user interface, wherein each of the one or more candidate attribute labels is corresponds to one of the following: a goal relevant to the task, a question related to the task, an answer to the question, and an insight that is different from the answer.

2. The system according to claim 1, further comprising: in response to the contextual element being selected at the user interface, determining a first attribute label, the first attribute label being associated with the contextual element such that the first attribute label is defined relative to the knowledge framework, the contextual element being based on the one or more source objects; providing the one or more candidate attribute labels as the auto-suggestion; and in response to the auto-suggestion being received as user input, enabling the user to input one or more other attribute labels such that the contextual element is associated with a label, the label being defined by at least the first attribute label, the second attribute label, the third attribute label, and the fourth attribute label, wherein the first attribute label corresponds to a goal relevant to the task, the second attribute label corresponds to a question related to the task, the third attribute label corresponds to an answer to the question, and the fourth attribute label corresponds to an insight that is different from the answer.

3. The system according to claim 2, further comprising: providing a user interface having an emergent labelling tool, the emergent labelling tool being configured to receive user input in response to the contextual element being selected, wherein the user input includes an attribute label, and wherein the attribute label is one of the following: the first attribute label, the second attribute label, the third attribute label, the fourth attribute label and/or one or more customized labels.

4. The system according to claim 3, wherein providing the auto-suggestion for the candidate second attribute label comprises: identifying a historical first attribute label from the plurality of first historical attribute labels based on a degree of similarity between respective associated contextual elements; and using a historical second attribute label that is associated with the historical contextual element as a candidate second attribute label of the auto-suggestion.

5. The system according to claim 4, wherein the knowledge framework includes historical first attribute labels received as user input from at least a first user, and wherein the auto suggestion is provided to a second user.

6. The system according to claim 3, wherein the auto-suggestion for the candidate second attribute label is based on one or both of the following: a historical second attribute label that is associated with a historical contextual element having a degree of similarity with the contextual element, the historical contextual element being associated with a historical first attribute label that is part of the knowledge framework; and a generative second attribute label, the generative second attribute label being generated by a method comprising: based on the contextual element as an input, obtaining an output from a natural language processing module; and iteratively comparing the output against a historical second attribute label using a generative adversarial network, wherein the historical second attribute label is selected based the first attribute label and the knowledge framework.

7. The system according to claim 6, wherein the wherein the knowledge framework includes historical first attribute labels received as user input from at least a first user, and wherein the auto-suggestion is provided to a second user.

8. The system according to claim 7, further comprising: generating a label based on a weighted ensemble of trained machine learning models, wherein a greater weightage is assigned to the model related to the historical fourth attribute labels relative to respective weightages assigned to any one of models of the historical first attribute labels, historical second attribute labels, and historical third attribute labels in the weighted collection of labels.

9. The system according to claim 7, wherein the emergent labelling tool further comprises: a manual interactive panel configured to receive manual user input; and an auto-suggestion interactive panel configured to provide the auto-suggestions.

10. The system according to claim 9, wherein the user interface further comprises a knowledge visualization tool configured to provide one or more auto-suggestions, wherein the one or more auto-suggestions are provided in response to a new source object being provided by one or more social media channel subscriptions linked to the task.

11. A method for deploying a software as a service according to the system as claimed in any one of claims 1 to 10.