Appendix 3: Types Of Cross References That Are Possible With Implicit References

Note: Need to retain old cataloging heading for machine matching purposes.

1. The local system could recognize an ampersand as "and" (including various languages) and vice versa. It may also be of value to maintain a list of common abbreviations and recognize them whether abbreviated or spelled out (e.g. search St. and get Saint and Sankt, and vice versa; search Mc and get Mac and vice versa), and perhaps to spell out numbers in English and other common languages as well. This seems relatively unproblematic once tables were in place for the desired words and symbols. If this were universally implemented for title searching as well, it would have the side benefit of eliminating the need for many of the title added entries which are mandated in bibliographic records. However, abbreviations, while useful for searching, may be problematic for batch authority control matching, and each abbreviation would need to be individually analyzed before this was implemented.
2. For rules regarding choice of entry element (inverted names and initials, multiple surnames, fuller name qualifiers), the local system could do by keyword searching. However, elimination of such references would be problematic for authority control vendors.
3. The local system could search from varying forms of fullness from less full to fuller (not counting initials and abbreviations). This could be done as a keyword search, but it may generate many false drops and make it unable to match an exact less-full form to a fuller form. This could also cover reference from full form of name to heading with initials but full form in q subfield. This would not be acceptable to batch authority control vendors, since too many headings would match multiple authority records, unless some other characteristics were also used for matching. This would inherently be much riskier than matching an explicit reference.
4. The local system could do automatic truncation to generate implicit reference from initials, shortened names (Rich to Richard, Ann to Anne, etc.), but this is much more likely to generate false drops in a large system., and would cause the same problems as no. 3 for authority control vendors.
5. The local system could employ spell-check or a "Nickname" list (like Rick to Richard, Anne to Ann), etc. to search variant spellings and less full forms, but again this introduces a false drop risk and risks for batch authority control matching.
6. The local system could search from general and specific names of conferences (as in less full to fuller names above), and also possibly from initials and acronyms to full name. This would have to be for any corporate body with those initials. This seems risky for both searching and for authority control processing.
7. The local system could search initials with periods and initialisms without, i.e., it should be able to search whether or not periods are present in the search or in the heading. Many searching algorithms already do this and it seems unproblematic. It could cause problems for authority control vendors should an initialism be the same as a word.
8. The local system could search from less full to fuller corporate names and varying entry element with a high risk of false drops and problematic authority control matches.
9. The local system could search from less full to fuller uniform titles with a high risk of false drops and problematic authority control matches.