Raldir / taxi

TAXI: a Taxonomy Induction Method based on Lexico-Syntactic Patterns, Substrings and Focused Crawling
Apache License 2.0
0 stars 0 forks source link

Make the prediction for all domain of the taxonomy induction #3

Open alexanderpanchenko opened 6 years ago

alexanderpanchenko commented 6 years ago

Make the prediction (in the similar fashion as #2 ) but for all domain of the taxonomy induction challenge. The candidates are from the vocabulary of the domain-specific words (e.g. food, science, etc. ~500–1000 words).

monsterwels commented 6 years ago

All predictions can be found here: https://www.dropbox.com/sh/gk1yqyi15a6v151/AAAPgxWR6FpJb-WCikXuGk4fa?dl=0

monsterwels commented 6 years ago
col0 col1 prediction prediction_score #paths
economics environment 1 0.9995301961898804 145
monkey environment 1 0.9991114735603333 1
balance environment 1 0.9988333582878113 55
lynx environment 1 0.9988332390785217 1
saltwater environment 1 0.9982742071151733 5
biotope environment 1 0.9978505373001099 1
plain environment 1 0.9972966313362122 22
estuary environment 1 0.996849000453949 18
shelf environment 1 0.9958900809288025 18
desert environment 1 0.9956746697425842 78
monsterwels commented 6 years ago
col0 col1 prediction prediction_score #paths
caramel food 1 0.8053730130195618 6
castella food 1 0.7408849596977234 1
pine food 1 0.7227337956428528 22
twist food 1 0.6970964670181274 12
sorbet food 1 0.6548916697502136 1
dew food 1 0.6468852758407593 6
cone food 1 0.645354688167572 5
juice food 1 0.6200062036514282 39
roe food 1 0.6140910387039185 7
harissa food 1 0.599799394607544 2
monsterwels commented 6 years ago
col0 col1 prediction prediction_score #paths
tribology science 0 0.2781708538532257 13
thermodynamics science 0 0.13678085803985596 38
sociolinguistics science 0 0.1182653158903122 1
syntax science 0 0.11715514212846756 18
econometrics science 0 0.0980641171336174 20
glaciology science 0 0.09769338369369507 16
isles science 0 0.09598596394062042 2
construction science 0 0.07844474166631699 310
bacteriology science 0 0.07381927967071533 27
pharmaceutics science 0 0.07086465507745743 2
monsterwels commented 6 years ago

And now path based

monsterwels commented 6 years ago
col0 col1 prediction prediction_score #paths
mechanical environment 1 0.999927282333374 2
wastage environment 1 0.9964731335639954 2
adaptation environment 1 0.9934678673744202 145
sensitive environment 1 0.9870176911354065 1
decontamination environment 1 0.9815698862075806 4
ships environment 1 0.9174445271492004 31
rodent environment 1 0.9081927537918091 5
desertification environment 1 0.8772224187850952 9
motor environment 1 0.813579797744751 9
volcanic environment 1 0.7295820116996765 1
monsterwels commented 6 years ago
col0 col1 prediction prediction_score #paths
kek food 1 1.0 2
lolly food 1 0.9998959302902222 2
teacake food 1 0.9993802309036255 2
nintendo food 1 0.9991999268531799 1
jeon food 1 0.9988155364990234 3
harissa food 1 0.9953378438949585 2
pho food 1 0.9928723573684692 2
platters food 1 0.9904733300209045 8
pepe food 1 0.9771053791046143 1
sancocho food 1 0.9601033329963684 4
monsterwels commented 6 years ago
col0 col1 prediction prediction_score #paths
biometry science 1 0.9999997615814209 3
interstellar science 1 0.9998096823692322 9
organization science 1 0.9997231364250183 585
organizations science 1 0.9996223449707031 159
film science 1 0.9995477795600891 670
neuroengineering science 1 0.9983460903167725 2
place science 1 0.99818354845047 429
esthetics science 1 0.9981158971786499 3
glaciology science 1 0.9965704679489136 16
pharmaceutics science 1 0.9963035583496094 2
alexanderpanchenko commented 6 years ago

important to confirm the unability of HypeNet to be used in taxonomy induction