Interactive Media & Services Data: Video, Streaming, Music Catalogs and Social Graphs

The three platform records anchor the slice. YouTube covers effectively all public videos with view counts, captions and region filters; Twitch serves realtime viewer counts, clips and creator analytics down to per-event pushes; Spotify exposes tracks, artists, albums, playlists and audiobooks including per-track audio features such as tempo and valence. All three arrive through Datadory with flattened, typed schemas.

Yes, but understand the mechanics: platform surfaces report current state, so history is accumulated rather than queried retroactively. Datadory delivers those accumulations on your cadence - hourly around launch windows, daily for standard research loops - so the series builds in your warehouse instead of inside someone else's quota window. Archived Wayback captures add before-and-after evidence for specific pages.

The Stanford SNAP Large Network Collection packages 80-plus graphs drawn from Facebook, Twitter/X, Google+, YouTube, Reddit, Twitch and LiveJournal, ranging from thousands of edges to com-Friendster's 1.8 billion. Files ship as edge lists and adjacency lists that load into any graph engine unmodified, which is why they remain the standard benchmark substrate for community detection.

Yes. TweetEval Benchmark fixes 200,785 tweets - roughly 14.2 MB - across seven tasks: irony, hate speech, offensive language, stance, emoji, emotion and sentiment, each with frozen train/validation/test splits. Fixed splits are the point: results become comparable to published baselines instead of re-partitioned approximations.

To 1996 for the web broadly: the Wayback Machine CDX index covers hundreds of billions of captures, and the earliest record observed in this slice is netflix.com in January 1999. Each capture record returns urlkey, timestamp, mimetype, statuscode, digest and length, filterable by URL pattern, date range and mimetype - enough to diff a competitor's page across a decade.

Yes. M-Lab contributes petabytes of speed-test and traceroute records continuous since February 2009, OONI Explorer adds hundreds of millions of censorship measurements from probes in nearly 200 countries since 2012, and APNIC Labs publishes daily IPv6, DNSSEC, RPKI and QUIC adoption series per country and ASN back to at least October 2013. Together they connect audience behavior to how the network actually performs.

Real rows, not screenshots: a video-channel extract limited to the creators and fields you specify, a week of viewer-count series at your chosen interval with the field dictionary attached, an edge-list slice sized to load in your environment, or a capture-history pull for named domains. Coverage statements and sample data are yours to keep whether or not you continue.