Collaborative filtering refers to the validation of data in the context of a community of users who are sharing data, the quality of which must be monitored. With regard to uScript, collaborative filtering involves the validation of each transcription by the users themselves, who will have more or less expertise in the area of manuscript transcription, rather than by an independent third party.
uScript will archive and rank all possible transcriptions for any word in a manuscript stored in its database. The “quality” of each word is determined by the users themselves, who implicitly “confirm” a previously-transcribed word when they do not offer an alternative transcription. Words left unchanged by a large number of users will be more credible, while the credibility for words for which users offer alternative transcriptions will drop. With collaborative filtering; therefore, the quality of a manuscript stored in uScript is assessed and recalculated each time it is accessed, whether or not the user makes any changes.