A controlled, real-life experiment designed to compare two variants of a system or a model, A and B.
In the context of Artificial Neural Networks, a function that takes in the weighted sum of all of the inputs from the previous layer and generates an output value to ignite the next layer.
A special case of Semi-Supervised Machine Learning in which a learning agent is able to interactively query an oracle (usually, a human annotator) to obtain labels at new data points.
An intelligent assistant that integrates into an existing customer support system. It helps improve the efficiency of customer service agents by quickly connecting them to the content they need to resolve customer issues.
Also known as surveillance or predictive agents, a type of intelligent software that continuously reports on computer equipment and system performance and malfunctions. They watch complex computer networks predicting when crashes or other defects may occur.
A strategic plan that outlines the steps and timeline for implementing AI technologies and initiatives within an organisation, considering factors such as budget, resources, and risk management.
A professional who provides expertise and guidance to businesses on leveraging artificial intelligence technologies to solve business problems, improve operations, and drive innovation.
The process of managing organisational change associated with the adoption and integration of AI technologies, including addressing employee concerns, training, and fostering a culture of AI adoption.
The ethical considerations and guidelines associated with the development, deployment, and use of AI systems, including issues related to fairness, accountability, transparency, and privacy.
The policies, procedures, and frameworks put in place to ensure responsible and effective use of AI technologies within an organisation, including guidelines for AI development, deployment, and monitoring.
An evaluation of an organisation's current AI capabilities, infrastructure, and processes to determine its level of AI maturity and identify areas for improvement.
The application of project management principles and practices to AI initiatives, including planning, resource allocation, risk management, and stakeholder communication.
The measurement of the financial or strategic benefits derived from AI investments, comparing the gains achieved against the costs incurred in implementing and maintaining AI systems.
The formulation and implementation of a plan to integrate AI technologies and solutions into a business's overall strategy, considering organisational goals, resources, and market opportunities.
The assessment of different AI technology providers or vendors to identify the most suitable solution based on factors such as functionality, scalability, cost, and support.
The combination of artificial intelligence (AI) and human insight (HI) that leverages the advantages and strengths of each to make them more efficient resources and to optimise customer service processes.
An unambiguous specification of a process describing how to solve a class of problems that can perform calculations, process data and automate reasoning.
Also known as Caller ID Spoofing, is an illegal activity used to defraud, harm, or obtain something of value from the recipient of a telephone call. This is accomplished by displaying a fraudulent Caller ID to the recipient.
A metadatum attached to a piece of data, typically provided by a human annotator.
A free, convenient way for customers to communicate with participating businesses. The communication uses the Messages app in iOS, macOS, watchOS, and iPadOS from different entry points like Wallet, Maps, or Search results.
A methodology used in Machine Learning to determine which one of several used models have the highest performance.
A methodology used in Machine Learning to determine which one of several used models have the highest performance.
An architecture composed of successive layers of simple connected units called artificial neurons interweaved with non-linear activation functions, which is vaguely reminiscent of the neurons in an animal brain.
A rule-based Machine Learning method for discovering interesting relations between variables in large data sets.
Also referred to as “async” messaging, describes any conversation where the participants are not concurrently active in the conversation via a messaging platform. This is great for the customer because they are able to start, pause, and resume a conversation at their convenience, and with the right tools a virtual or human agent can pick up the conversation right where they left off.
A type of Artificial Neural Network used to produce efficient representations of data in an unsupervised and non-linear manner, typically to reduce dimensionality.
A subfield of Computational Linguistics interested in methods that enables the recognition and translation of spoken language into text by computers.
A common call centre metric representing the average duration of a customer interaction. It starts with the customer’s initiation of a call and includes all talk and hold time until the termination of the call.
A method used to train Artificial Neural Networks to compute a gradient that is needed in the calculation of the network’s weights.
The set of examples used in one gradient update of model training.
A famous theorem used by statisticians to describe the probability of an event based on prior knowledge of conditions that might be related to an occurrence.
Inductive Bias: the set of assumptions that the learner uses when predicting outputs given inputs that have not been encountered yet.
A conflict arising when data scientists try to simultaneously minimise bias and variance, that prevents supervised algorithms from generalising beyond their training set.
Extremely large and complex data sets that require advanced tools and techniques to store, process, and analyse.
A Machine Learning ensemble meta-algorithm for primarily reducing bias and variance in supervised learning, and a family of Machine Learning algorithms that convert weak learners to strong ones.
The smallest (rectangular) box fully containing a set of points or an object.
It refers to a collection of platforms and tools utilised by companies and consumers for effective communication. The predominant form of messaging channel is SMS or text messaging, commonly employed for sharing shipping notifications, marketing offers, promotional campaigns, and appointment reminders. With the advancement of technology, consumers are increasingly adopting newer messaging channels such as Apple Business Chat and Google Business Messages. These channels enable seamless interaction with brands at any time, from any location, and across various devices.
The use of AI technologies, such as robotic process automation (RPA) or intelligent process automation (IPA), to automate repetitive tasks, streamline workflows, and improve operational efficiency.
The contracting out of non-primary business activities to third-party providers. BPO services include payroll, human resources (HR), accounting and customer/call centre activities. It may also be referred to as Information Technology Enabled Services (ITES).
The electronic circuitry within a computer that carries out the instructions of a computer program by performing the basic arithmetic, logical, control and input/output operations specified by the instructions.
A computer program or an AI designed to interact with human users through conversation.
The task of approximating a mapping function from input variables to discrete output variables, or, by extension, a class of Machine Learning algorithms that determine the classes to which specific instances belong.
In Machine Learning, the unsupervised task of grouping a set of objects so that objects within the same group (called a cluster) are more “similar” to each other than they are to those in other groups.
A potential issue arising from the fact that a system cannot infer anything for users or items for which it has not gathered a sufficient amount of information yet.
A method used in the context of recommender systems to make predictions about the interests of a user by collecting preferences from a larger group of users.
Also known as social tagging or folksonomy. Enables users to apply public tags to online items. This activity is typically done to make items easier to find at a later date.
The field of Machine Learning that studies how to gain high-level understanding from images or videos.
Refers to a program’s ability to divide a task into parts that can be processed independently of each other but at the same time. This allows a single task to be executed out of order, and still have the overall result remain the same as if they were executed in a defined order, allowing the overall task to be completed faster.
A type of interval estimate that is likely to contain the true value of an unknown population parameter. The interval is associated with a confidence level that quantifies the level of confidence of this parameter being in the interval.
The tendency to search for, interpret, favour, and recall information in a way that confirms one’s own beliefs or hypotheses while giving disproportionately less attention to information that contradicts it.
Manages customer inquiries via voice calls, email, messaging apps, and other data applications. In comparison, a Call Center only handles incoming and outgoing voice calls.
A cloud-based customer experience solution that allows a company to utilise a Contact Center provider's software. A CCaaS business model allows a company to purchase only the technology they need and reduce the need for internal support.
The bidirectional transfer of information between two parties where both are aware of the relational, environmental, and cultural context of the exchange. In simple terms, it is a conversation between two parties where both sides are completely aware of all the aspects of the conversation.
A human worker providing annotations on the Appen data annotation platform.
Refers to evaluation of a visitor-agent conversation transcript and patterns for policy compliance, conversation efficiencies, and fraud detection/prevention.
A conversational AI system launched in an ad setting typically aimed to drive brand or product awareness.
Systems employ AI technology to manage interactions between chatbots, acting as a bridge between businesses and individuals. These systems consist of two parts: utterances (what the customer says) and responses generated by the chatbot. The main aim of conversational AI is to engage customers within the same conversation, guiding them throughout their entire experience.
These interactions primarily focus on obtaining information and enabling transactions, such as payments or purchases (also known as conversational commerce). Conversational AI systems can be deployed through various platforms, including messaging apps, voice assistants, and chatbots. By leveraging contextual cues and user data, these systems create personalised customer experiences that align with the specific communication channel, industry, and customer journey.
Refers to customer support, questions and answers, personalised recommendations, reviews, and purchases happening within a messaging app or format. In this form of engagement, the consumer may be interacting with a human representative, a chatbot, or a mix of both.
A design language used to create a conversation flow and establish an underlying logic. It integrates several disciplines to create a language based on human conversation, including voice/user interface and interaction design, plus visual, motion, audio design, and UX writing.
A class of Deep, Feed-Forward Artificial Neural Networks, often used in Computer Vision.
A metric illustrating the cost per conversation with a customer/prospect. It is usually calculated as the cost of operating a contact centre during a set period of time divided by the total number of conversations realised during that time.
A KPI measurement on how much each contact costs your call centre. It is a key part of cost-benefit analyses and tracks the wage and operating costs associated with each time an agent picks up the phone or sends an email.
A collection of processes designed to evaluate how the results of a predictive model will generalise to new data sets.
- k-fold Cross-Validation
- Leave-p-out Cross-Validation
A single-item metric that measures the effort necessary from a customer to get an issue resolved, a request fulfilled, a product purchased/returned, or a question answered.
Refers to the emotional connection between a customer and a brand/company. It is demonstrated by the sum total of consumer interactions that suggest increased brand loyalty, purchase increase, and voluntary promotion of a brand.
A customer’s interaction with an organisation, company, or brand. A positive experience helps strengthen customer loyalty, brand affinity, and increase revenue. It also generates the potential for positive referrals to new customers.
The sum-total experiences that a customer has with a company or brand. It includes each and every touchpoint from first contact through final purchase, customer service, and after-sale support.
A customer’s attachment to a company, a product, or a service and their perceived value of that company and their dedication to that brand above all others in the marketplace.
A measurement of how “happy” a customer is with a company's products, services, and capabilities. Customer satisfaction information is collected through surveys and ratings and can help a company determine improvements and changes to its products and services with the potential to strengthen its competitive advantage.
It is the most essential ingredient to all Machine Learning and Artificial Intelligence projects.
Data processed in a way that it becomes ingestible by a Machine Learning algorithm and, if in the case of Supervised Machine Learning, labelled data; data after it has been processed on the Appen data annotation platform.
Raw, unprocessed data. Textual data is a perfect example of unstructured data because it is not formatted into specific features.
The process of adding new information derived from both internal and external sources to a data set, typically through annotation.
The framework and processes are implemented to ensure the quality, integrity, and privacy of data throughout its lifecycle, including data collection, storage, sharing, and usage.
An interdisciplinary field that combines statistics, programming, and domain knowledge to extract insights and knowledge from data.
A comprehensive plan that outlines how an organisation collects, manages, analyses, and utilises data to drive AI initiatives and make informed business decisions.
A category of Supervised Machine Learning algorithms where the data is iteratively split in respect to a given parameter or criteria.
A chess-playing computer developed by IBM, better known for being the first computer chess-playing system to win both a chess game and a chess match against a reigning world champion under regular time controls.
A broader family of Machine Learning methods based on learning data representations, as opposed to task-specific algorithms. Deep Learning can be supervised, semi-supervised or unsupervised.
A neural network with a level of complexity that exceeds two layers. Deep neural networks use sophisticated mathematical modelling to process data in complex ways.
A solution or a group of solutions that provide autonomous consumer activity on a website or intranet. This approach is typically used for simple support activities such as managing contracts, requesting quotes, or finding the answer to a customer’s question.
The application of digital technologies to create new or modified business processes, culture, and customer experiences to meet changing business and market needs.
Dimensionality Reduction: the process of reducing the number of random variables under consideration by obtaining a set of principal variables. Also see Feature Selection.
Curse of Dimensionality: phenomena that arise when analysing and organising data in high-dimensional spaces due to the fact that the more the number of dimensions increases, the sparser the amount of available data becomes.
An interactive voice response format where callers are presented with options, such as yes or no questions or pre-programmed responses, to move the interaction forward.
One instance of some mathematical structure contained within another instance, such as a group that is a subgroup.
In Statistics and Machine Learning, ensemble methods use multiple learning algorithms to obtain better predictive performance that could be obtained from any of the constituent learning algorithms alone. Unlike a statistical ensemble in statistical mechanics, which is usually infinite, a machine learning ensemble consists of only a concrete finite set of alternative models but typically allows for a much more flexible structure to exist among those alternatives.
The average amount of information conveyed by a stochastic source of data.
In the context of training Deep Learning models, one pass of the full training data set.
A measure of a model’s accuracy considering both the precision and the recall to compute the score. More specifically, the F-Score is the harmonic average of the precision and recall, where it reaches its maximal value at 1 (perfect precision and recall) and minimum at 0.
An error due to the fact a result did not reject the null hypothesis when it should have.
An error due to the fact a result did reject the null hypothesis when it shouldn’t have.
A variable that is used as an input to a model.
The process of selecting, transforming, and creating relevant features from raw data to improve the performance of machine learning models.
An Artificial Neural Network wherein connections between the neurons do not go backward or form a cycle.
An important contact centre metric and a critical element of customer relationship management. A contact centre's ability to resolve customer issues the first time a customer calls, requiring no further follow-up.
A set of activities conducted to determine whether fraudulent activity has occurred.
Frequently Asked Questions (FAQs) refer to a compilation of popular questions that users or customers often ask. It offers clear and concise answers to address their queries and alleviate any concerns they may have. By presenting information in a question-and-answer format, it provides a convenient way for users to find the information they need quickly. This not only saves time but also reduces the necessity for individuals to contact customer support or search extensively for answers. It is an effective tool for simplifying complex topics and making them easily understandable to users.
A principle stating that whenever the input data is flawed, it will lead to misleading results and produces nonsensical output, a.k.a. “garbage”.
A mobile conversational channel that combines entry points on Google Maps, search, and brand websites to create a rich, asynchronous messaging experience to engage customers and drive business results.
A regulation in EU law on data protection and privacy for all individuals within the European Union aiming to give control to citizens and residents over their personal data.
A class of Artificial Intelligence algorithms used in Unsupervised Machine Learning, implemented as the combination of two Neural Networks competing with each other in a zero-sum game framework.
A search heuristic inspired by the Theory of Evolution that reflects the process of natural selection where the fittest individuals are selected to produce offspring of the following generation.
The distribution of contact centre agents across multiple geographic areas, including work-from-home options, to ensure business continuity in case of a location shut down. Ensuring your data centre locations are distributed globally and are able to operate in a standalone capacity and deliver redundancy required to maintain business critical services.
A labour market characterised by short-term and freelance contracts as opposed to permanent jobs. Also referred to as the sharing economy or the collaborative economy, and often in reference to platforms such as Airbnb and Uber.
A specialised electronic circuit designed to rapidly manipulate and alter memory to accelerate the rendering of images thanks to its parallel processing architecture, which allows it to perform multiple calculations simultaneously.
A piece of information obtained through direct observation as opposed to inference.
A heuristic is a problem-solving technique or mental shortcut that enables efficient decision-making and problem-solving. It is a practical strategy used when the best solution is uncertain or time-consuming to determine.
A practical problem-solving approach, offers efficient and intuitive methods for finding solutions. Instead of relying solely on logical reasoning, it uses mental shortcuts based on past experiences. This technique enables quicker decision-making and more effective problem-solving. Heuristics enable practical quick decision-making when working with complex datasets.
Human-in-the-Loop (HITL) is a process that involves combining human intelligence with automated systems to achieve more accurate and reliable results. In HITL, humans play a vital role in tasks that require cognitive abilities, while machines handle repetitive or computational tasks. This collaborative approach ensures that human expertise is utilised where it matters most, enhancing the overall efficiency and effectiveness of the system.
A hyperparameter is a parameter that is set before training a machine learning model and cannot be learned from the data. It determines the behaviour and performance of the model during training and influences the final results.
Hyperparameter tuning refers to the process of selecting the best values for these hyperparameters. By adjusting the hyperparameters, we can optimise the model's performance and improve its accuracy.
The problem in Computer Vision is determining whether an image contains some specific object, feature, or activity.
A large visual dataset made of 14 million URLs of hand-annotated images organised in twenty-thousand (20,000) different categories, designed for use in visual object recognition research.
The process of making predictions by applying a trained model to new, unlabeled instances.
The area of Computer Science studying the process of searching for information in a document, searching for documents themselves, and also searching for metadata that describes data, and for databases of texts, images or sounds.
Detecting and identifying a user’s intent from their written and spoken communication. It forms an important part of dialogue modelling for both AI and voice agents.
Intent models are powerful tools used in natural language processing (NLP) and machine learning to understand the purpose or intention behind a user's input. These models analyse textual data and predict the underlying intent of the user, enabling businesses to provide more accurate and relevant responses.
Intent models play a crucial role in various applications, such as chatbots, virtual assistants, and customer support systems. By accurately determining user intent, these models help automate interactions and deliver personalised experiences. For example, an intent model can identify whether a user's query is seeking information, making a reservation, or requesting support.
The result of leveraging pattern data to determine intent before a customer takes action.
Leverages intent discovery from contextual scenarios to route conversations to the agent most suited to address the situation. For example, cues from the utterances of an increasingly frustrated customer might route the conversation to a live agent.
Communication or direct involvement with a live agent or a bot by phone, messaging, email, or in person.
Interactive Voice Response (IVR) systems, also known as IVRS, are automated telephony systems that enable businesses to interact with customers through voice prompts and keypad inputs. They handle high call volumes, providing pre-recorded information and services. IVR systems are widely used in customer support, call centres, and other phone-based services. They streamline customer inquiries, offer self-service options, and route calls efficiently.
Used to describe the interactions a customer has with a company on all available channels: telephone, web, social media, in-store, and email, for example. Often broken into the 2 distinct paths, sales and customer service.
The process of actively working toward creating positive customer experiences, rather than relying on it to happen organically. The end goal is a seamless, interactive customer journey across existing systems, channels, and touchpoints.
The conscious process of defining, structuring, retaining, and sharing the knowledge and experience within an organisation. The main goal of knowledge management is to improve an organisation's efficiency and keep knowledge capital within the company.
A series of neurons in an Artificial Neural Network that process a set of input features, or, by extension, the output of those neurons. Hidden Layer: a layer of neurons whose outputs are connected to the inputs of other neurons, therefore not directly visible as a network output.
A new direction within the field of Machine Learning investigating how algorithms can change the way they generalise by analysing their own learning process and improving on it.
The application of Machine Learning to the construction of ranking models for Information Retrieval systems.
A scalar value used by the gradient descent algorithm at each iteration of the training phase of an Artificial Neural Network to multiply with the gradient.
Also referred to as live help or live support, it allows a customer to communicate with customer service representatives in real time. Rather than force a customer to speak to a representative on the phone, visitors can have a live interaction with agents.
The inverse of the sigmoidal “logistic” function used in mathematics, especially in statistics.
A variation of Recurrent Neural Network proposed as a solution to the vanishing gradient problem.
The subfield of Artificial Intelligence that often uses statistical techniques to give computers the ability to “learn”, i.e., progressively improve performance on a specific task, with data, without being explicitly programmed.
Machine Learning Lifecycle Management (MLLM) is the holistic approach to efficiently managing and improving machine learning models at every stage of their lifecycle. This includes gathering data, creating models, implementing them, overseeing performance, and ensuring ongoing maintenance.
A subfield of computational linguistics that studies the use of software to translate text or speech from one language to another.
Processes and systems put into place to ensure a unified articulation of messages and policies in a company.
A model is an abstract representation of what a Machine Learning system has learned from the training data during the training process.
In computer science, a Model Abstraction Layer is the generalisation of a conceptual model or algorithm, not specific to a particular implementation. It is a way to hide the implementation details of deep functionality. The separation facilitates interoperability and platform independence. This is how peripheral devices can work with different computers or operating systems.
An approximate methodology that uses repeated random sampling in order to generate synthetic simulated data.
A subfield of Machine Learning that exploits similarities and differences across tasks in order to solve multiple tasks are at the same time.
A marketing strategy that offers customers a variety of ways to purchase a product. A multichannel strategy covers purchases from a mix of in-store, website, telephone, mail orders, interactive television, catalogue ordering and comparison-shopping sites. These channels tend to be siloed and often treated as separate business units that have little interaction with one another. (also see Omnichannel)
A subfield of Machine Learning aiming to interpret multimodal signals together and build models that can process and relate information from multiple types of data.
A family of simple probabilistic classifiers based on applying Bayes’ theorem with strong independence assumptions between the features.
A sub-task of Information Extraction that seeks to identify and classify named entities in text into predetermined categories such as the names, locations, parts-of-speech, etc.
The area of Artificial Intelligence that studies the interactions between computers and human languages, in particular how to process and analyze large amounts of natural language data.
A metric used to assess customer loyalty for a brand, product, or service. It is often used as part of a customer relationship management strategy because the metric is relatively easy to calculate.
Used to define a multichannel approach to sales, marketing or customer service that focuses on providing seamless customer experiences—whether the consumer engages via phone, website chat, messaging or mobile app. It is also referred to as Omnichannel Retail or Omnichannel Commerce.
The conversion of images of printed, handwritten or typed text into a machine-friendly textual format.
The selection of the best element (with regard to some criterion) from some set of available alternatives.
A message routed from a client or an application and delivered to the end user's mobile phone. It may also be referred to as a Mobile Terminated message (MT) and signifies that the endpoint of the message is a mobile phone.
The fact that a model unknowingly identified patterns in the noise and assumed those represented the underlying structure; the production of a model that corresponds too closely to a particular set of data, and therefore fails to generalise well to unseen observations.
An area of Machine Learning focusing on the (supervised or unsupervised) recognition of patterns in the data.
Using the knowledge, a company has about a customer to tailor interactions and experiences for that customer using technology. Often used as part of their marketing and customer relationship management (CRM) strategy.
The process of reducing a matrix generated by a convolutional layer to a smaller matrix.
Any piece of information that can be used on its own or in combination with some other information in order to identify a particular individual.
The number of correct positive results divided by the number of all positive results returned by a classifier.
The inferred output of a trained model provided with an input instance.
The process of transforming raw data into a more understandable format.
A model, or the component of a model, that have been preliminary trained, generally using another data set.
A process that uses an orthogonal transformation to convert a set of observations of possibly correlated variables into a set of linearly uncorrelated variables called principal components.
The probability distribution that would represent the preexisting beliefs about a specific quantity before new evidence is considered.
An ensemble learning method that operates by constructing a multitude of decision trees at training time and outputting a combined version (such as the mean or the mode) of the results of each individual tree.
The fraction of all relevant samples that are correctly classified as positive.
A unit employing the rectifier function as an activation function.
A class of Artificial Neural Network where connections between neurons form a directed graph along a sequence, allowing it to exhibit dynamic temporal behaviour for a time sequence and to use their internal state (memory) to process sequential signals.
A set of statistical processes for estimating the relationships among variables.
An area of machine learning that focuses on the suitability of a particular action to maximise the reward in a particular situation. In the absence of a training dataset, the system learns from experience.
A protocol between both mobile operators and phones replacing SMS messages with a richer text-message system and multimedia support, all via the data network.
Refers to software that can be programmed to do basic tasks across applications—just like human workers would do. RPA software is designed to reduce the burden of simple, repetitive tasks for employees.
A model that spans all the possible topics that is typically present at the root node or very start of the conversation. It can also be globally available in the conversation to support topic changes mid-conversation. A journey-level model is a more specific model tailored to the specific node in a journey-specific path with topics and interaction points that are specific to a given journey and the end user's place inside the journey.
The process of determining whether a piece of writing is positive, negative, or neutral. This helps data analysts gauge public opinion, conduct nuanced market research, monitor brand and product reputation, and better understand customer experiences.
Also known as opinion mining or emotion AI, it is a subdiscipline of machine learning and natural language processing that tries to identify the emotions within a text or speech sample.
A case assignment strategy used in support centres to assign incoming cases to the most suitable agent, rather than simply choosing the next available agent.
Also referred to as Smart Reply or Smart Compose, it is an AI technology that searches the information available on a platform (email, text, message apps) and offers three response options to choose from based on most frequent (or relevant) answers.
Insights that can be used as indicators regarding interest levels in a company or product obtained by collecting social media data. Information such as who customers are following and/or engaging with including relevant key influencers, or what they are talking about regarding relevant topics.
Speech recognition is the ability of a computer to process and understand spoken words. It is also known as automatic speech recognition (ASR), computer speech recognition, or speech-to-text (STT). Speech recognition systems use computer algorithms to convert spoken words into text.
The ability for a computer to recognize and identify a person by their spoken voice. The application of this technology allows a system to limit access and control various services by voice.
The act of using an automated speech recognizer (ASR) to turn human spoken utterances into transcribed digital text representations.
In statistics, an empirical distribution function is the distribution function associated with the empirical measure of a sample. This cumulative distribution function is a step function that jumps up by 1/n at each of the n data points. Its value at any specified value of the measured variable is the fraction of observations of the measured variable that are less than or equal to the specified value.
A form of machine learning (ML) that “learns” a particular function by way of mapping an input to an output based on sample input-outputs. A supervised learning algorithm analyses new data in comparison to its example data to produce an inferred function, which can then be used to further map new examples.
A class of discriminative classifiers formally defined by a separating hyperplane, where for each provided labelled training data point, the algorithm outputs an optimal hyperplane which categorises new examples.
A live conversation with a clear start and end that typically has faster customer query response times due to the live and present interaction of both parties. The conversation has a defined beginning and end, and in the context of customer support, the agent solves the customer's issue(s) at the end of the conversation.
Data generated artificially when real data cannot be collected in sufficient amounts, or when original data doesn’t meet certain requirements.
An open-source library, popular among the Machine Learning community, for data flow programming across a range of tasks. It is a symbolic maths library and is also used for machine learning applications such as neural networks.
In the context of Supervised Machine Learning, the process of assessing the final performance of a model using hold-out data.
Testing Data: The subset of available data that a data scientist selected for the testing phase of the development of a model.
A sequence of data points recorded at specific times and indexed accordingly to their order of occurrence.
A category of Unsupervised Machine Learning algorithms that uses clustering to find hidden structures in textual data and interpret them as topics.
In the context of Supervised Machine Learning, the construction of algorithms that can learn from and make predictions from data.
Training Data: The subset of available data that a data scientist selected for the training phase of the development of a model.
Monitoring transactions in a way that provides proof of compliance and counters the risks of money laundering and terrorist financing.
An area of Machine Learning that focuses on using knowledge gained to solve a specific problem and apply this knowledge to a different but related problem.
A form of speech synthesis used to create a spoken version of the text found in an electronic document.
A test developed by Alan Turing to evaluate a machine’s ability to exhibit intelligent behavior equivalent to that of a human. The test consists of having the machine chat with a human. If a human evaluator witnessing the conversation from outside the room where the test takes place can’t reliably tell the machine apart, the machine is said to have passed the Turing test.
In statistical hypothesis testing, a type I error is the rejection of a null hypothesis that is actually true. This means that the researcher concludes that there is a statistically significant difference between the two groups when, in reality, there is no difference. Type I errors are also known as false positives.
In statistical hypothesis testing, a type II error is the failure to reject a null hypothesis that is actually false. This is also known as a false negative. A type II error can occur when the sample size is too small, the effect size is too small, or the variability of the data is too large.
A range of values likely to enclose the true value.
The fact that a Machine Learning algorithm fails to capture the underlying structure of the data properly, typically because the model is either not sophisticated enough or not appropriate for the task at hand; opposite of Overfitting.
It is a type of machine learning that allows algorithms to find patterns in unlabeled data. This type of learning is different from supervised learning, which requires labeled data to train the algorithm.
Utterance is a continuous piece of speech, often beginning and ending with a clear pause. It can be anything from a single word to a sentence, or even a paragraph. Utterances can be spoken or written, and they can be used to express a wide range of ideas and emotions.
The process of using hold-out data in order to evaluate the performance of a trained model; by opposition to the testing phase which is used for the final assessment of the model’s performance, the validation phase is used to determine if any iterative modification needs to be made to the model.
A dreaded difficulty and major obstacle to recurrent net performance that data scientists face when training Artificial Neural Networks with gradient-based learning methods and backpropagation, due to the neural network’s weights receiving an update proportional to the partial derivative of the error function with respect to the current weight in each iteration of training.
An error due to sensitivity to small fluctuations in the training set computed as the expectation of the squared deviation of a random variable from its mean.
An automated or computerized customer service. Virtual agents can provide relevant advice and information to a customer and sustain a relevant conversation via a voice channel, web site, messaging or mobile apps.
Also referred to as Voice Recognition, Speaker Recognition, Voice Printing, and Voice Authentication. Used to verify the identity of a speaker. Voice biometrics map a speaker's unique voice characteristics and later uses the map for identification. A user typically provides one or more audio samples which the system analyses to create a unique voiceprint for the speaker.
A cloud-based, artificial intelligence-enabled smart adviser software, often used in conjunction with standalone smart speakers (such as Amazon Echo®), can also be integrated with other devices/equipment.
When an employee works from their place of residence, rather than from the office. In 2020, the COVID-19 pandemic forced many companies to create WFH policies and allowed their employees to work from home full-time or when it is most convenient for them.
Various methods of workplace surveillance to gather information about the activities and locations of staff members. It is often used to improve productivity and protect corporate resources.