Glossary

Definitions for the terms you’ll see across these docs and in the product. Platform terms come first, followed by the underlying concepts and techniques.

​

Platform terms

​

Concepts & techniques

​

End-to-End Solution

An end-to-end solution refers to a product, service, or system that covers all stages of a process from start to finish without the need for additional software or interventions. In the context of data science, an end-to-end solution means that the entire process from data collection to using AI applications to represent your data, can all be done within a single framework. End-to-end solutions allow Corvic to be more efficient and allow the user to move from one phase to the next without the need for multiple tools or manual interventions.

​

Foreign Key and Primary Key

A primary key is a unique identifier for each record in a database table, ensuring that no two rows share the same value and preventing repeated entries, while a foreign key is a column that establishes a link between two tables by referencing the primary key of another table. Together, these feature types facilitate relationships between tables, enabling efficient data retrieval. This helps to keep the data organized and connected in relational databases. If you would like to learn more about Primary Key and Foreign Key, those links will navigate you to articles with more information.

​

Generative AI

Generative AI (GenAI) is a type of artificial intelligence that uses algorithms to produce new content based on existing data. GenAI can generate text, images, audio, and video by identifying patterns and structures in large datasets. It relies on training data to understand context, style, and content, allowing it to create outputs that resemble human-made creations. Given inputs and parameters, GenAI is able to produce information and responses quickly and accurately. The applications span various fields, including marketing, entertainment, and design.

​

Graph AI

Graph AI refers to a type of artificial intelligence that uses graph-based structures to represent and analyze data. In simple terms, it treats information as a network of interconnected points or “nodes” and connects them. This approach helps in understanding complex relationships and patterns within data, allowing you to gain deeper insights into your data. Graph AI leverages the way relationships between data points influence each other to make smarter, more informed decisions. For example, in a tech company, Graph AI can analyze how different systems, applications, or users are connected and at the same time detect unusual patterns or behaviors, such as outliers that might appear with the data you upload.

​

Image Spaces

Image Spaces represent visual data based on their content and visual features. They generate embeddings for images by transforming them into points in a high-dimensional space where visually similar images are placed closer together. This method helps in capturing the complex visual patterns, textures, and structures present in images. For example, in computer vision tasks, Image Spaces can differentiate between various objects, scenes, or styles, improving image classification, similarity search, and even generation tasks. By capturing the visual essence of images in a compact form, Image Spaces enable efficient processing and analysis of large-scale image datasets.

​

Multi-Modal Embeddings

Multi-Modal Embeddings are representations that combine information from different types of data such as text, images, graphs, etc. Embeddings can also capture relationships and interactions between different types of data, for example, a multimodal embedding can link a text description of an image to the image itself. If you would like even more information on what embeddings are, this link will navigate you to an article on embeddings.

​

Parquet File Format

The Parquet file format is a columnar storage file format designed for efficient data processing particularly in large data sets. It optimizes storage by compressing and encoding data and is particularly beneficial for data science, as Parquet allows for efficient retrieval of large datasets, especially when only specific columns are needed. Additionally, Parquet works with data processing frameworks like Apache Spark and Hadoop, making it an essential choice for data scientists to analyze large volumes of structured and semi-structured data. If you would like even more information on what Parquet File Format is, this link will navigate you to a video with more information.

​

Predictive AI

Predictive AI is a type of artificial intelligence that analyzes historical data to forecast future outcomes and trends. It uses statistical algorithms and machine learning models to identify patterns in existing datasets and extrapolate likely future scenarios. It processes both structured and unstructured data to make informed predictions about events, behaviors, and performance metrics. The applications span various fields, including finance, healthcare, and logistics, where it helps organizations make data-driven decisions and prepare for future developments.

​

Relational Spaces

Relational Spaces refer to frameworks used in artificial intelligence to model and analyze the relationships between different entities or concepts. They map data into a space where the relationships between entities are as important as the entities themselves. This approach helps in understanding how different elements interact and influence each other. For instance, in a recommendation system, Relational Spaces can represent users and products in a way that captures how user preferences and behaviors relate to various products, enabling more accurate recommendations based on these interconnections.

​

Semantic Spaces

Semantic Spaces are models that represent data based on its meanings and contextual relationships. In simple terms, they place data points, such as words or concepts, in a high-dimensional space where their meanings and relationships are captured. This method helps in understanding the context and subtleties of the data. For example, in natural language processing, Semantic Spaces can differentiate between the meanings of similar words and improve text understanding and generation by capturing the contextual nuances of language.

​

Tabular Spaces

Tabular Spaces convert rows of numerical and categorical variables into points within a multi-dimensional space. In such spaces, similar data points are positioned closer together, revealing underlying patterns and relationships. These models excel at capturing complex interactions between various data types, such as age, income, and preferences. By doing so, Tabular Spaces enhance the performance of analytical tasks like classification, regression, and clustering. For instance, in customer analytics, they can uncover hidden correlations between demographics and purchasing behaviors, leading to more accurate predictions and deeper insights.

​

UMAP

UMAP (Uniform Manifold Approximation and Projection) is a technique for reducing the dimensionality of high-dimensional data while preserving its structure. In simple terms, it maps complex data into a lower-dimensional space, making it easier to visualize and analyze. This approach helps in identifying patterns, clusters, and relationships within the data. For example, in data visualization, UMAP can transform a dataset with many features into two or three dimensions, allowing users to explore and interpret the underlying structure and groupings more intuitively.