SSF Slot PER_NAME, ORG_NAME, FAC_NAME

[Wallace]

Hi, John,

I am confused about Slot:  PER_NAME, ORG_NAME, FAC_NAME    in YAML file.

In "trec-kba-2014-07-11-truth-data/trec-kba-2014-assessor-guidelines.pdf", it list the slots of PER, ORG ,FAC that will use in SSF Task. 

On one hand,  it doesn't contain the PER_NAME, ORG_NAME, FAC_NAME.    On the other hand, we cannot use all the slots in YAML file.

How does the slot "PER/FAC/ORG_NAME" use?

Thanks very much.

Wallace

[John R. Frank]

*_NAME is the list of alternate names for the entity, which arevery useful
for many IR purposes, so having slot filling algorithms that accurately
identify alternate names is valuable.

In SSF, the task is to start with this input:

* target_id
* canonical_name
* external_profiles
* all relevance rating assertions (before and after the time cutoff)

and generate output records that include this info:

* target_id
* stream_id
* slot_name
* offset(s)
* confidence score


Does that answer your question?


jrf

[Wallace]

Does that mean we should find the "*_NAME" slot in SSF Task?  

But the slot "*_NAME" is not in Slot list defined in "truthdata/guideline.pdf".

 

在 2014年8月27日星期三UTC+8下午7时44分25秒，Wallace写道：

[John R. Frank]

> Does that mean we should find the "*_NAME" slot in SSF Task?  But the
> slot "*_NAME" is not in Slot list defined in "truthdata/guideline.pdf".

The target slots for KBA are defined in:

trec-kba-2014-07-11-ccr-and-ssf.slot_names.json

See the README.txt circulated in the data tarball.

Some of the slots didn't need special guidance, so our only described in
the KBP Guidelines or ACE guidelines, which are here:

https://github.com/trec-kba/streamcorpus/blob/master/if/


jrf