What are some good search engines to do scientific research?

How do research papers on information retrieval from the major search engines get published, when none of their results can be verified?

• Search engines companies like Google, Yahoo & Microsoft publish a lot of work at top conferences using private datasets which can't be used by other researchers to verify their results. Is it ethical to publish those results?

• Answer:

  This is more of an issue of peer review than anything else. Then again, in biology there often will be use of "in-house" strategies that often will be difficult to reproduce. Technically the data is reproducible. If you could make your own private dataset, you theoretically could verify their results. A good example in the medical world would be: Company makes a library, company does high-throughput screen, company identifies compound. Likewise, search companiese are doing the same thing. They build the dataset, mine it, and determine hits. At the end of the day, its not the data that matter but rather the algorithm or the "hit" and I think that is what truly is being presented in the publications.