Daylight: The Luminoso product that analyzes natural language text documents and produces insights about those documents. Daylight uses Luminoso QuickLearn technology to learn specific meanings of words and phrases in a dataset. Daylight identifies concepts that are most meaningful within a project and identifies conceptual matches.
Compass: The Luminoso product that automatically analyzes text samples as they arrive in real time and classifies them based on labeled examples or defined topics that users provide.
Document: A document is a single unstructured natural language text sample and optional metadata that you upload to Daylight or classify using Compass. In a CSV file, a document is one row. Most Daylight projects are made up of thousands of documents.
Metadata: Any additional data that accompanies the unstructured text on a document. Metadata could include information like date, number, star rating, or demographic. You can also include qualitative labels like “male,” or “Massachusetts.” After uploading metadata, sort and filter your data to drill down to specific insights.
Project: The basic unit of analysis in Luminoso Daylight. A project is built out of a set of unstructured text samples and associated metadata that make up documents. Each project is built based on words in context, so concepts and metrics for each Daylight project are unique to that project.
Filter: You can use metadata to filter documents in Daylight and view specific subsets of documents. For example, you might create a filter based on location and age to see only documents from 45- to 65-year-olds from Massachusetts or Rhode Island.
Concept: A concept is a term or phrase that appears in your dataset. It can be a single word like "amazing" or "smelly," or a phrase, like "love my dog" or "not helpful.”
Saved concept: Save a concept to access it easily in Daylight. Once you save a concept, associate a color with it in the Galaxy feature and view its exact and conceptual matches. You can rename a saved concept, though renaming a concept won’t change the way it interacts with your project.
Exact match: An exact match is an identical match to a concept you select. For example, if you analyzed the concept "app," exact matches might include the concept "app," "App", or "apps" but not closely related concepts like "tablet" or "download.”
Conceptual match: A conceptual match is a close but not identical match to a concept you select. Since using only exact matches might skip important information, Luminoso also provides a powerful conceptual matching tool. For example, if you select the concept "delicious," conceptual matches might include "yummy" and "tasty."
Association score: Association scores measure how related two concepts are on a scale of -1.00 to 1.00. These numbers represent the weakest and the strongest possible relationships. A concept has an association score of 1.00 with itself. A score of 0 indicates that two concepts are only as associated with one another in a project as they would be randomly across the whole language.
Prevalence: If a word or a phrase appears more frequently in a project than it does in the language as a whole, it is prevalent. For example, in a consumer electronics project, the word “wifi” may occur 1000 times more frequently than it does across the English language. This indicates the word’s prevalence in the project.
Highlights: The Daylight feature where you can quickly view high-level insights about your project. Use Highlights to orient yourself to your project and start navigating the questions you can answer using Daylight.
Volume: Use Daylight’s Volume feature to gain quick insight into how concepts are represented across the dataset. Volume automatically quantifies how frequently concepts are referenced in the project using the exact and conceptual matching identified by QuickLearn.
Sentiment: Use Daylight’s Sentiment feature to evaluate whether documents are positive, negative, or neutral and measure the amount of times concepts appear in positive or negative documents. Sentiment helps you locate documents where a concept appeared in a positive or negative context.
Drivers: Find terms that are prevalent and reveal a significant impact on a measurable aspect of your data in Daylight’s Drivers feature. For example, in a restaurant review project where “scores” are star ratings, the concept “long line” might be a negative driver. A driver’s impact represents the difference between the average score for all documents and the average score for documents that contain the concept you selected. If “long line” has an impact of -0.6 stars, reviews containing “long line” or its conceptual matches are an average of 0.6 stars lower.
Galaxy: In the Galaxy feature, view strong conceptual relationships are in the context of your entire project. The size of a concept in the visualization indicates how relevant it is in the project. Related concepts cluster together by theme. You can also use Galaxy to manage concepts.
QuickLearn: The transfer learning technology that Luminoso uses to learn from unstructured natural language text samples. QuickLearn combines a background space trained on ConceptNet and a variety of freely available text and domain text. QuickLearn then integrates the text documents that you want to analyze and creates a new space of word embeddings specifically tuned to your data. QuickLearn understands specific meanings of in-domain words contextually and general meanings of common words through the background space.