Re: jmdict: question about ent_seq

Paul Blay wrote:
<andersson.johnny@xxxxxxxxx> wrote ...

What exactly is the purpose of the ent_seq element in jmdict?
According to the DTD, it's "A unique numeric sequence number for each
entry". Does that mean that I can use that number to unambigously
identify a particular word across different versions of the file? That
is, if I have a function in my software to create bookmarks to certain
entries, can I I use ent_seq to find the exact same entry even after
updating to a never release of jmdict - no need to store kanji/kana or
other means of identifying the entry?

Yes, with some provisos. First there are 'TempSUB' entries that users
have submitted but haven't been processed yet. They aren't released
in the standard jmdict but they have temporary numbers that get changed
later.

Not quite. Those entries aren't in JMdict, and don't have sequence numbers.

Second entries may be merged, deleted or have the headwords and
or readings added or corrected. So the entry you link to may disappear
or no longer be the same as when you linked to it.

The database from which JMdict is generated has a record of deleted
entries, but I haven't carried those records into the XML version. I'd
probably have to introduce an extra top-level entity, or pehaps have
another file of the sequence numbers of deleted records.

At present there is not a significant body of software using the
XML format - most people seem to stick with the old EDICT version. If
more people were using JMdict, and if the sequence numbers of deleted
records became an issue, I could certainly do something about it.

--
Jim Breen http://www.csse.monash.edu.au/~jwb/
Clayton School of Information Technology,
Monash University, VIC 3800, Australia
ジム・ブリーン@モナシュ大学
.

Relevant Pages
Loading