> ## Documentation Index
> Fetch the complete documentation index at: https://vicewire.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Entity Relevance Signals

> How ViceWire records whether the company is central to the article or only mentioned in passing.

Company Relevance Signals are article-level fields inside `media_item`. They help distinguish articles that are **primarily about the company** from articles where the company is only mentioned in passing, as context, as background noise, or in peripheral page elements such as related-story tabs, modules, or other adjacent content that is not central to the article itself.

These are not event-level fields because they describe the source article as a whole, not any one extracted event object. The `media_item` is always scoped to the security defined in `entity`, and any events extracted from that article are interpreted in relation to that same security.

## Why this layer exists

A company can appear in an article in very different ways.

For example, some articles are clearly centered on Apple. Others only mention Apple as a benchmark, competitor, historical reference, or side note. That distinction matters because a passing mention is not as useful as an article where Apple is the main subject.

These signals help answer a few simple questions:

* is the company the main focus of the article?
* is the company mentioned in the title?
* how often is the company mentioned?
* how early does the company appear in the article?

## Fields

<ResponseField name="focus_of_article">
  `focus_of_article` is a simple article-level focus flag.

  It is set to `true` when the company is the main focus of the article, and `false` when the company is only mentioned in passing or as background context.
</ResponseField>

<ResponseField name="mentioned_in_title">
  `mentioned_in_title` records whether the company is mentioned in the article title.

  This is a simple prominence signal. A title mention often indicates that the company is central to the piece, although it should be interpreted alongside the other relevance fields.
</ResponseField>

<ResponseField name="entity_count">
  `entity_count` records how many times the company is mentioned in the article.

  This is a rough prominence signal rather than a standalone relevance judgment. Higher counts often indicate greater centrality, but repetition alone does not guarantee that the company is the true focus of the piece.
</ResponseField>

<ResponseField name="first_sid_mentioned">
  `first_sid_mentioned` records the first sentence ID (`sid`) in which the company appears.
</ResponseField>

<Info>
  These fields are best read together. `focus_of_article` is the primary relevance judgment, while `mentioned_in_title`, `entity_count`, and `first_sid_mentioned` are supporting prominence signals.
</Info>

## How to interpret them

These signals are designed to support filtering and prioritization.

For example:

* an article with `focus_of_article = true`, `mentioned_in_title = true`, and an early `first_sid_mentioned` is more likely to be directly about the company
* an article with `focus_of_article = false`, a low `entity_count`, and a late `first_sid_mentioned` is more likely to be incidental or background mention coverage

## Where they sit in the payload

These signals form part of the `media_item` group because they describe article-level relevance rather than event-level meaning.

```json theme={null}
{
  "media_item": {
    "entity_count": 33,
    "mentioned_in_title": true,
    "focus_of_article": true,
    "first_sid_mentioned": 1
  }
}
```

## The core idea

Company Relevance Signals help answer a simple but important question: **is this article really about the company, or is the company only part of the background context**?

By keeping those signals explicit at the article level, ViceWire makes it easier to filter, rank, and interpret coverage before or alongside the event-level output.