Ask HN: what type of sources should I use for my novel unsearch engine

Yes you read that right we are building an <i>un-search</i> engine. We are very different from existing search engines. However we are in a dilemma to choose data sources to build out our MVP. Our search is most effective and value clear when multiple heterogeneous data are indexed (obviously we have anything but a Google data farm indexing power). We are trying to pick a domain or source that is manageable scale for a small startup, and yet shows the ability to "weave" interdomain knowledge. The obvious ones are wikipedia, wikibooks etc. but would the community have any suggestions on on other sources? news perhaps? Thanks in advance!