This table does not contain actual author data besides e-mail addresses. Rather, it is a relationship table between DOCUMENT and PERSON. Each record in this table describes a "Person was an author of Document" relationship. That being said, both the PERSON and AUTHORS tables are being referred to when mentioning authors in this article.
Authors are parsed primarily out of the "AU" ISI field, which all valid ISI datasets should contain, but they can also be parsed out of References. The remainder of this article will speak as if parsing out of the "AU" field.
If provided, full author name data is parsed out of the "AF" field, which is expected to contain as many entries as the "AU" field. Entries in both fields are matched by the order they were listed; there is no error checking/correction if the orders mismatch.
If provided, the e-mail address data is parsed out of the "EM" field, which is expected to contain either as many entries as the "AU" field or one entry. If the number of entries in both fields match, the entries are matched by the order they were listed (similarly to full author names). Otherwise, the first (or only) e-mail address is used for all authors.
The abbreviated name fields in the PERSON table (FAMILY_NAME, FIRST_INITIAL, MIDDLE_INITIAL, and UNSPLIT_NAME) are parsed from the "AU" field. This article describes how the "AU" field is parsed. Note: Even when the "AF" field is supplied, the FAMILY_NAME is parsed out of the "AU" field.
The full name fields in the PERSON table (ADDITIONAL_NAME, FULL_NAME, and PERSONAL_NAME) are parsed from the "AF" field. This article describes how the "AF" field is parsed.
This table contains the following fields:
In the various built-in extractions, when Authors are specified in the "AU" field, they are considered to be inner authors. When they are specified as part of reference strings and are not merged, they are considered to be outer authors.
See DOCUMENT, PERSON, REFERENCE, How Abbreviated Names are Parsed, and How Full Names are Parsed for additional details.