mulangonando / impact-of-kg-context-on-ed Goto Github PK

There are some serious discrepancies between the code and the paper.
Below is the image describing the inputs into the roberta model

However, after running the scripts, The discrepancies are:

1. There is no surface mention in the training inputs. instead the Wikidata id is input, which makes no sense to the model
2. Entity context is jumbled up pretty bad, the predicates and the objects are not in order at all.
3. Padding token is wrong for roberta (1 - )

Below is a sample training input:

['<s>',
 'ĠQ',
 '5',
 '34',
 '153',
 '</s>',
 'Ġfantasy',
 'Ġnovelist',
 'ĠDavid',
 'ĠGem',
 'm',
 'ell',
 '.',
 'ĠAchilles',
 'Ġis',
 'Ġfeatured',
 'Ġheavily',
 'Ġin',
 'Ġthe',
 'Ġnovel',
 'ĠThe',
 'ĠFire',
 'brand',
 'Ġby',
 'ĠMarion',
 'ĠZimmer',
 'ĠBradley',
 '.',
 'ĠThe',
 'Ġcomic',
 'Ġbook',
 'Ġhero',
 'ĠCaptain',
 'ĠMarvel',
 'Ġis',
 'Ġendowed',
 'Ġwith',
 'Ġthe',
 'Ġcourage',
 'Ġof',
 'ĠAchilles',
 ',',
 'Ġas',
 'Ġwell',
 '</s>',
 'Ġcountry',
 'Ġof',
 'Ġcitizenship',
 'Ġinspired',
 'Ġby',
 'ĠZach',
 'ary',
 'ĠLevi',
 'ĠUnited',
 'ĠStates',
 'Ġof',
 'ĠAmerica',
 'Ġperformer',
 'ĠSuperman',
 '</s>',
 '<s>',
 '<s>',
...

Loving the paper.

Is the code for the context builder released in this repo?
I am unable to find how the training / testing data is generated.

Context builder referring to the module in the image in your paper below.

Thank you.

Hi,

Thanks for releasing the code. Can you provide more detail on the KG-context for the 2-hop triples? I cannot fully understand it from the paper since only an example of the 1-hop triple in Figure 2 was demonstrated. Also, I think in your given dataset with context, only 1-hop counts were provided (correct me if I am wrong). It would be great if you can provide a few examples.

Moreover, can you also inform what was the SPARQL query that was run?

Thanks.

Hi,

I understand that one of the key contributions of the paper is to represent the triples in KG context in the natural language. However, looking through the data in the Roberta folder, the KG context seems to be jumbled up.

For instance: Q8242
Sample expected context:
6 Q8242 Slovak literature is the literature of Slovakia. History. Middle Ages. The first monuments of literature in present-day Slovakia are described by source literary science studied by Jewish Encyclopedia of Brockhaus and Efron literary criticism subclass of art 1

Sample actual context:
6 Q8242 Slovak literature is the literature of Slovakia. History. Middle Ages. The first monuments of literature in present-day Slovakia are described by source Jewish Encyclopedia of Brockhaus and Efron studied by literary criticism subclass of art 1

Should this be a concern?

Thank you