Haro (1956) used multiply exposed large scale plates to search for emission line galaxies. He identified a number of compact galaxies with strong emission lines. Zwicky (1971) envisioned that galaxies would evolve into highly concentrated densities as for neutron stars and produced several lists of ``compact galaxies'' candidates selected form the Palomar sky survey. A nearly contemporary line of research was opened by Markarian (1967) who focused on ultraviolet excess galaxies at the Byurakan observatory. The underlying assumption was completely opposite however since V.A. Ambartsumian had the view that galaxy nuclei originated from explosive events.
Many metal-poor galaxies known at present undergo enhanced star formation event. This type of activity has obviously favoured their discovery (a metal-poor galaxy not necessarily needs to be an active one!). Following the early discovery of two bona fide metal-poor galaxies by Searle and Sargent (1972) that indeed turned out to experience an active star formation episode many astronomers have embarked in the building of large samples of such objects. Fundamentally, with the aim to study their statistical importance in the Universe as a function of time, their relation to large structures and their contribution to the luminosity function.
One way to select star-forming galaxies is to rely on their ultraviolet excess and emission line spectrum. This property is not completely unambiguous since it is shared to some extent by active galactic nuclei (AGN). There are complications of course, mainly because of the effect of dust that reduces the UV flux and re-radiate the Lyman continuum photons in the far infrared (IR excess in IRAS 60 micron band). Follow-up observations need to distinguish the full population of starbursts from the AGN with non thermal activity.