SUBJECT: Re : how to talk about a distinction in text mining ? &NAME , In the terminology of the &NAME &NAME &NAME ( &NAME ) program ( &WEBSITE &NAME &NUM relations are termed ' explicit ' , and &NAME &NUM relations are termed ' implicit ' . See the site documentation for more details . &NAME In text mining , I find it useful to distinguish between these &NUM ways in which text can carry information about real world entities : &NAME &NUM . The information about the entity is present as a natural language utterance in the text , e.g. ' Mrs. ' &NAME lives in &NAME . ' &NAME &NUM . The textual record is linked to the person , and that link is critical for using the text as a source of information about the real world entity . For instance , Mrs. &NAME in an email to a social newsgroup use various vocabulary , make references to locations , etc. that would lead one to believe that the author of the message lives in &NAME . The distinction is not sharp , and the same piece of text can obviously carry information in both ways . Linguistically perhaps the difference just boils down to resolving the personal pronoun ' &NUM ' using the ' Sender : ' field of the email rather than a name in the text . In practice , however , I find the &NUM to be very different . In particular , &NAME &NUM is much more susceptible to text classification approaches . One assumes the whole document is ' about ' the person to which it is linked , and you classify the person by classifying the document . In &NAME &NUM , however , you have to be much more serious about the &NAME , since perhaps only that single sentence talks about the entity you 're interested in , and may do some in a way where the linguistic structure makes a lot of difference . Anyway , I 'm trying to think of a good pair of terms to talk about &NAME &NUM vs. &NAME &NUM . ' &NAME ' vs. ' &NAME ' ? ' Third person vs. First person ' ? Any suggestions ? Cheers , &NAME &NAME &NAME &NAME &WEBSITE &NAME on text categorization / filtering / search / mining / ( your need here )