Glossary¶
Ecosystem Terms¶
Term |
Definition |
|---|---|
Snippet |
Short news report, processed post, or extraction from documents, such as articles, audio, or video files. |
Snippet ID |
A unique identifier assigned to each Snippet for tracking and referencing. |
Document |
Documents are either a post, article, podcast, or video file. They can contain multiple reports with independent information depending on the type of document. For example, large documents, like articles, podcasts, and videos, may contain multiple reports, while a post on social media is often a single report. |
Document URL |
URL links to the document where the snippet was originally generated. |
Date/Time |
Timestamp for when the report was published. The API is in UTC and Platform uses your desktop’s local browser time. |
Relevancy |
Probability, or confidence score, indicating the contextual relevance of a snippet to sports betting or market-changing events. |
Entity |
Key people, places, or things. For example, a player or team. |
News Type |
Predicted probability that a report is related to any of our pre-defined types of news, including: Injury, Lineup, Practice, Suspension, Trade, Contract, Draft, Hiring, Firing, Performance, Other |
Primary News Type 1 |
The news type with the highest probability. |
Primary News Type 2 |
The news type with the second highest probability. |
Source |
The place origin of where document was published. |
Source Type |
Types of documents, such as social media, articles, podcast, or video. |
Extracted URL |
URL extracted from a snippet, often a link to a complementary article, press conference, or player interview. |
Author |
The person that wrote and published the document, such as an article. For example, the ‘By line’ of an article. |
Reporter |
The person cited for publishing the information contained in a snippet. For example, Pierre LeBrun, NHL Insider |
Reporter Type |
Classification of reporters based on their type of coverage, either National or Local. National reporters cover the entire league, while local reporters often cover a single team. |
Reporter Company |
The company that the reporter primarily works for, such as The Athletic or Washington Post. |
Primary Team |
The team that the reporter has mentioned most often in their reports. |
Parent ID |
A field in the snippet object that references the snippet_id of an earlier, related snippet. When null, indicates the snippet is an original source or contains unique information. |
Snippet Linking |
The process of identifying relationships between snippets, specifically determining if a new snippet contains the same core information as a previously ingested snippet. |
Technical Terms¶
Term |
Definition |
|---|---|
Semantic Search |
Search methodology that understands the contextual meaning of queries rather than simple keyword matching, allowing for more natural language searching and finding conceptually similar content even when exact terms differ. |
Bi-Encoder |
A neural network architecture used in our snippet linking process that efficiently computes semantic similarity between text snippets to quickly retrieve potential matching candidates. |
Cross-Encoder |
A neural network architecture used in our snippet linking process that performs detailed comparison between pairs of snippets to determine if they contain the same core information, providing high-precision matching. |