You might view (my) problem as an embedding for words (and its fragments) driven by valued statements (those you discard), and then inverting this (learned encoder) into a language model. Thus when describing an object it would be possible to chose better words (lexical choice in natural language generation).
On Mon, Oct 2, 2017 at 5:00 PM, fn@imm.dtu.dk wrote:
I have done some work on converting Wikidata items and properties to a low-dimensional representation (graph embedding).
A webservice with a "most-similar" functionality based on computation in the low-dimensional space is running from https://tools.wmflabs.org/wemb edder/most-similar/
A query may look like:
https://tools.wmflabs.org/wembedder/most-similar/Q20#language=en
It is based on a simple Gensim model https://github.com/fnielsen/wembedder and could probably be improved.
It is described in http://www2.imm.dtu.dk/pubdb/v iews/edoc_download.php/7011/pdf/imm7011.pdf
It is not embedding statements but rather individual items.
There is general research on graph embedding. I have added some of the scientific articles to Wikidata. You can see them with Scholia:
https://tools.wmflabs.org/scholia/topic/Q32081746
best regards Finn Årup Nielsen http://people.compute.dtu.dk/faan/
On 09/27/2017 02:14 PM, John Erling Blad wrote:
The most important thing for my problem would be to encode quantity and geopos. The test case is lake sizes to encode proper localized descriptions.
Unless someone already have a working solution I would encode this as sparse logarithmic vectors, probably also with log of pairwise differences.
Encoding of qualifiers is interesting, but would require encoding of a topic map, and that adds an additional layer of complexity.
How to encode the values are not so much the problem, but avoiding reimplementing this yet another time… ;)
On Wed, Sep 27, 2017 at 1:23 PM, Thomas Pellissier Tanon < thomas@pellissier-tanon.fr mailto:thomas@pellissier-tanon.fr> wrote:
Just an idea of a very sparse but hopefully not so bad encoding (I have not actually tested it). NB: I am going to use a lot the terms defined in the glossary [1]. A value could be encoded by a vector: - for entity ids it is a vector V that have the dimension of the number of existing entities such that V[q] = 1 if, and only if, it is the entity q and V[q] = 0 if not. - for time : a vector with year, month, day, hours, minutes, seconds, is_precision_year, is_precision_month, ..., is_gregorian, is_julian (or something similar) - for geo coordinates latitude, longitude, is_earth, is_moon... - string/language strings: an encoding depending on your use case ... Example : To encode "Q2" you would have the vector {0,1,0....} To encode the year 2000 you would have {2000,0..., is_precision_decade = 0,is_precision_year=1,is_precision_month=0,...,is_gregorian=true,...} To encode a snak you build a big vector by concatenating the vector of the value if it is P1, if it is P2... (you use the property datatype to pick a good vector shape) + you add two cells per property to encode is_novalue, is_somevalue. To encode "P31: Q5" you would have a vector V = {0,....,0,0,0,0,1,0,....} with 1 only forV[P31_offset + Q5_offset]
To encode a claim you could concatenate the main snak vector + the qualifiers vectors that is the merge of the snak vector for all qualifiers (i.e. you build the vector for all snak and you sum them) such that the qualifier vectors encode all qualifiers at the same time. it allows to check that a qualifiers is set just by picking the right cell in the vector. But it will do bad things if there are two qualifiers with the same property and having a datatype like time or geocoordinates. But I don't think it really a problem. Example: to encode the claim with "P31: Q5" main snak and qualifiers "P42: Q42, P42: Q44" we would have a vector V such that V[P31_offset + Q5_offset] = 1, V[qualifiers_offset + P42_offset + Q42_offset] = 1 and V[qualifiers_offset + P42_offset + Q44_offset] = 1 and 0elsewhere.
I am not sure how to encode statements references (merge all of them and encode it just like the qualifiers vector is maybe a first step but is bad if we have multiple references). For the rank you just need 3 booleans is_preferred, is_normal and is_deprecated. Cheers, Thomas [1] https://www.wikidata.org/wiki/Wikidata:Glossary <https://www.wikidata.org/wiki/Wikidata:Glossary> > Le 27 sept. 2017 à 12:41, John Erling Blad <jeblad@gmail.commailto:jeblad@gmail.com> a écrit : > > Is there anyone that has done any work on how to encode statements as features for neural nets? I'm mostly interested in sparse encoders for online training of live networks. > > > _______________________________________________ > Wikidata mailing list > Wikidata@lists.wikimedia.org mailto:Wikidata@lists.wikimedia.org > https://lists.wikimedia.org/mailman/listinfo/wikidata https://lists.wikimedia.org/mailman/listinfo/wikidata
_______________________________________________ Wikidata mailing list Wikidata@lists.wikimedia.org <mailto:Wikidata@lists.wikimedia.org> https://lists.wikimedia.org/mailman/listinfo/wikidata <https://lists.wikimedia.org/mailman/listinfo/wikidata>
Wikidata mailing list Wikidata@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata
Wikidata mailing list Wikidata@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata