This page is a summary of information about the Lojban Dictionary: It's history, current form, and future design.
Dictionary Software and Data
- jbovlaste is the Lojban dictionary.
- la sutysisku, a slick interface for la jbovlaste database with some algorithms improving search relevance.
- Lojban Online Dictionary Query. The jbovlaste search engine.
- Dictionaries, Glossers and parsers contains links to software for using a Lojban Dictionary.
- Word Lists contain dictionary information in a variety of formats.
Several pages on the wiki discuss the Lojban Dictionary. Those are collected here:
- The Lojban Dictionary
- Great Dictionary Problem
- dictionary notes from 2001
- BPFK Section: Dictionary Preface
- Old Draft Dictionary
- Mini-dictionary To-do
- Mini-dictionary paradigms
This section is an exploration of the issue started in this discussion about the dictionary backend.
What use cases will use Lojban dictionary data?
What end-use will this data put put to?
- A print version of the Lojban dictionary
- jbovlaste, the online dictionary. This includes editing and voting on dictionary entries
- Flash card definitions, which are more concise.
- Interactive tools, like glossers and parsers
- Ideally, jbofihe could use this data instead of having its own version.
- Similarly, the list of rafsi for jvocuhadju could be extracted herefrom
What sorts of things should a Lojbanic dictionary store, ideally?
- Are any of these use cases unsuitable for a dictionary? Are there use cases we haven't thought of that are suitable?
- Can we separate a definition from its grammatical context? Are we treating Lojban gismu too much like a verb or noun in the way we handle them now?
- Should we use some manner of spatial visualization? (e.g., cpacuvisualization)
- Versioning: it would be really really nice to be able to have BPFK proposals reference "checkpoint 123 in the dictionary software". A possibility here: have an "official word" checkbox, and a button that says "checkpoint all the official words, give the checkpoint a number", and the ability to diff successive checkpoints. Or possibly have the checkpoint button do everything, but be able to filter out the diffs to just show the official stuff, or not. Then it could be of more general use: could also have a "this word is worth printing" button, and than any given official print run of the dictionary would be all those words, from checkpoint #123 or whatever.
- Related: You can't edit a definition with the "officially approved" and "official word" checkboxes. You can fork it, and then there's a button that outputs a diff between the fork and the officialdata version. An admin could then move the "officially approved" checkbox around.
What storage format is going to work for all of these use cases?
- Each use of the dictionary data needs to view it in a different way. Can we design a format that can be shared between all users of the data?
- Since the form of a dictionary entry is often unstructured text, What would a dictionary definition look like that supports all of our use cases without duplicating the definitions by changing their form? (e.g., a brief definition for a study card, a full definition for a print dictionary, and an archive discussion for the online dictionary?)
What prior art is there?
- Dictionaries are not a new problem, how do other people deal with this?
- Is Lojban fundamentally different because of its formal grammar? Is our thinking on this problem influenced by working with languages that don't have a formal grammar: Is a dictionary a compromise that we have a better solution for?
gismu, lujvo, and selma'o have different storage requirements. The proposal below assumes they will be storted in a database, and describe the storage schema.
<tab class="wikitable">Field Description gismu the gismu this row defines lang the language this row is in definition the definition of the gismu x1 gloss for x1 x2 gloss for x2, NULL if there is no x2 x3 gloss for x3, NULL if there is no x3 x4 gloss for x4, NULL if there is no x4 x5 gloss for x5, NULL if there is no x5 xN gloss for xN, NULL if there is no xN </tab>
There is one entry in this table for each language/gismu.
<tab class="wikitable"> Field Description lujvo the lujvo this row defines lang the language this row is in definition the definition of the lujvo x1 gloss for x1 x2 gloss for x2, NULL if there is no x2 x3 gloss for x3, NULL if there is no x3 x4 gloss for x4, NULL if there is no x4 x5 gloss for x5, NULL if there is no x5 xN gloss for xN, NULL if there is no xN </tab>
<tab class="wikitable">Field Description ]] lujvo the lujvo this row defines component the component gismu or lujvo that is one part of this lujvo lang x1 The position this component's x1 place appears in the lujvo. NULL if it does not appear. x2 The position this component's x2 place appears in the lujvo. NULL if it does not appear. x3 The position this component's x3 place appears in the lujvo. NULL if it does not appear. x4 The position this component's x4 place appears in the lujvo. NULL if it does not appear. x5 The position this component's x5 place appears in the lujvo. NULL if it does not appear. xN The position this component's xN place appears in the lujvo. NULL if it does not appear. </tab>
There is an entry in this table for each component of the lujvo.
<tab class="wikitable">Field Description selma'o the selma'o this row defines lang the language this row is in class the selma'o class of this selma'o gloss the definition of this selma'o </tab>
How will this format be used in each of the use cases we've identified?
A print version of the Lojban dictionary
Assuming we're printing the English language translation dictionary, the gismu table will be rendered as the string:
$gismu: $x1 $definition $x2 $x3 $x4 $x5 $xN
jbovlaste, the online dictionary.
To be done.
To be done.