How does APITube ensure data quality?#

APITube handles data quality in the open: instead of hiding a black-box score, it attaches machine-readable quality signals to every article — the source’s OPR authority score, sentiment for the headline and the body separately, and the publisher’s political bias — and gives you filter switches to enforce your own bar. You are never forced to trust a single “quality” number; you can read the signals on each article and decide what passes.

Those signals arrive inline. A normal search on /v1/news/everything already carries the enrichment, so you do not call a separate quality endpoint to get it.

How does APITube express data quality?#

APITube treats quality as two things you can act on: signals and filters.

• Signals are fields on every article object — the source’s source.rankings.opr, the sentiment block, the author, and the publisher source.bias. They let you judge and audit each result.
• Filters are request parameters that turn those signals into a hard cutoff — for example is_high_quality=1, is_premium_source=1, or an OPR floor with source.rank.opr.min.

Both come from the same processing step, so a search returns judged, filterable articles rather than raw scraped text. Send your key as a header (X-API-Key: YOUR_API_KEY) or as the api_key query parameter.

How does source authority (OPR) work as a quality signal?#

Every source is scored with OPR, an authority score from 0 to 10 for the publishing domain. APITube attaches that score to each article under source.rankings.opr, and you can set a floor or ceiling with source.rank.opr.min and source.rank.opr.max:

curl "https://api.apitube.io/v1/news/everything?source.rank.opr.min=6&api_key=YOUR_API_KEY"

OPR rates the publisher’s standing, not whether one story is accurate — every article from the same domain shares the same OPR. For the full scale and its meaning, see what is the OPR score.

What one-flag quality filters can I turn on?#

Instead of listing domains by hand, APITube ships ready-made switches built on the OPR bar. Each one is a precise, documented rule:

• is_high_quality=1 keeps an article only when all four conditions hold: it is not flagged as a duplicate, its source OPR is 5 or higher, it carries at least one image, and it has a named author.
• is_verified_source=1 keeps articles from OPR-5+ publishers that are not flagged as duplicates.
• is_premium_source=1 keeps only the top tier, OPR 6 or higher.

Because is_high_quality also checks the article (image, byline), it is stricter than a source-only flag — a top publisher’s story is still dropped if it has no image or no author. Pick the right switch in how to filter for high-quality, verified sources.

How does APITube flag misleading headlines?#

Data quality is also about honesty between the headline and the body, so APITube scores both separately and exposes the mismatch. is_clickbait=1 keeps articles where the headline polarity differs from the body polarity and the headline’s sentiment score is strongly charged (absolute value above 0.5) — an alarmist title over a calm story. Flip it to is_clickbait=0 to exclude those articles from clean results.

For finer control, sentiment_gap.min and sentiment_gap.max measure the raw distance between title and body sentiment on a 0 to 2 scale, so you can keep only articles whose headline matches their content. A full walkthrough is in how to detect and filter clickbait.

How can I audit why an article passed the filter?#

Every article reports the signals the filters checked, so quality is verifiable, not implied. In each object you get the source’s source.rankings.opr, three-level sentiment (overall, title, body), the publisher source.bias (left, center, or right), and the author. Pull a sample without any quality flag, read those fields, then decide whether the built-in OPR-5 floor is enough or you want to raise it with source.rank.opr.min. Nothing about a result’s quality is hidden from the response.

Common Questions#

• Does a quality flag guarantee the article is accurate?
• What is the fastest way to get clean results?
• Does bias filtering remove low-quality sources?

Does a quality flag guarantee the article is accurate?#

No. OPR measures publisher authority, and the sentiment and clickbait signals measure tone, not factual accuracy. A high-OPR, non-clickbait article is a stronger, cleaner result — but the score does not verify the claims inside it. Treat these signals as reputation and consistency checks, and pair them with your own verification when accuracy is critical.

What is the fastest way to get clean results?#

Add is_high_quality=1 to your request. It applies all four checks at once — non-duplicate, OPR 5 or higher, has an image, has a named author — so you get presentable, well-sourced articles from a single parameter without maintaining a source list.

Does bias filtering remove low-quality sources?#

No. source.bias is a transparency label (left, center, or right), independent of authority. Filtering or excluding a bias does not raise quality — use OPR and the quality flags for that, and use source.bias only to balance or narrow the political leaning of your feed. For the exact field names on every article, see the official News API parameters reference.