Terms last and last year further qualify the particular day, but when the year is explicitly stated as in Cinco de Mayo 2000, we annotate Cinco de Mayo as ^ and annotate 2000 as z , because within this example, the date term refers to a complete date Might 5, 2000. We do annotate time with the day applying the label d , but we also think that it’s too general to link to the patient for re-identification. Since we usually do not classify it under the Date category, we don’t annotate time periods within every day as W (e.g., noon-4:30pm); instead, we use label d to annotate noon and 4:30pm, separately. 3.6. Telecommunication and Alphanumeric IdentifiersTelecommunication identifiers would be the most straightforward along with the least ambiguous identifiers given that they’re well defined engineering objects. On the 18 individual identifiers defined by the HIPAA Privacy Rule, five of them are telecommunication identifiers to be de-identified: phone numbers, fax numbers, electronic email addresses, web universal resource locators (URLs), and Internet protocol (IP) address numbers. As new telecommunication modes and media emerge, new telecommunication identifiers (e.g., Twitter usernames such PubMed ID:http://www.ncbi.nlm.nih.gov/pubmed/21307382 as BarackObama) seem, but they also are covered by the final (18th) catch-identified. Numeric and Alphanumeric Identifiers consist of four labels: D Z E ,W E ,, Z , and . The initial two are very precise identifiers denoting medical record and protocol numbers,respectively. Healthcare record number is one of the 18 HIPAA identifiers. Since protocol numbers are very crucial entities for clinical researchers, that are the intended users of NLM Scrubber, we annotated them separately. We make use of the label , Z for all other alphanumeric identifiers issued by overall health care and FT011 manufacturer insurance coverage providers uniquely to the patient; e.g., hospital account number, well being program beneficiary quantity and lab specimen quantity. D Z E ,W E and , Z are nearly always connected together with the patient only in a handful of circumstances we did observe mentions of such identifiers for the relatives. We use the label for all other numeric and alphanumeric identifiers that happen to be not issued by the provider, like those 5 identifiers defined by the Privacy Rule: social safety quantity, account numbers, certificate license numbers, car identifiers, and device identifiers. Note for hospital account numbers we use the label , Z . Occasionally, names of some lab materials and experimental drugs may perhaps contain some numbers (e.g., drug 123-ABC or instrument QRS-40). We do not annotate such well being data, as they’re neither special for the patient nor personal identifiers. 3.7. Personally Identifying ContextSo far, we discussed how we annotate entities that have been talked about within the HIPAA Privacy Rule in conjunction with several other closely associated entities, a number of which is usually PII in particular contexts. We’re aware of the reality that due to the intricacies of natural languages, it truly is feasible to specify a context in which the individual might be identified indirectly such that no labels we discussed so far could be acceptable to make use of. In these instances, we label the tokens with W, denoting Personally Identifying Context. reporting with label K W and Tahrir Square with W W , because the latter would present context so specific that as well as the occupation information would likely recognize the person directly. In inside the military, but the deployment to Iraq will not be an occupation equipment could possibly be deployed to Iraq also as other forms of personnel such as reporters c.