/LING

Update: February 24, 2023 The new version of Termout.org is now online, so this web site is now obsolete and will soon be dismantled.

Lista de candidatos sometidos a examen:
1) precision (*)
(*) Términos presentes en el nuestro glosario de lingüística

---

1) Candidate: precision

Is in goldstandard

---

1
paper corpusSignosTxtLongLines177 - : BLEU is basically a precision metric: it measures which n-grams in the candidate answer appear in the references . In the case of scoring student's answers, we both want the answer to be correct and complete. Therefore, we have modified this metric to calculate as well the percentage of the references that is covered by the student's answer. To do that, the Brevity Factor is substituted by a Modified Brevity Penalty (MBP) calculated in the following way: for each reference, calculate the percentage of n-grams that is covered by the candidate text, and, next, we add up all those percentages. Figure 4 shows an example in which 5% of the first reference, 10% of the second one and 20% of the third one appears in the student's answer. Therefore, we can assume that 35% of a complete answer is covered by the candidate text. The results using this MBP clearly outperform those obtained using the original algorithm.

---

2
paper corpusSignosTxtLongLines455 - : * F-Measure represents the weighted harmonic mean of precision and recall and is determined by: (2 x Precision x Recall ) / (Precision + Recall)

Evaluando al candidato precision:

1) covered: 3
3) candidate: 3
4) recall: 3 (*)

precision
Lengua: eng
Frec: 20
Docs: 10
Nombre propio: 1 / 20 = 5%
Coocurrencias con glosario: 1
Puntaje: 1.801 = (1 + (1+3.32192809488736) / (1+4.39231742277876)));
Candidato aceptado

Referencias bibliográficas encontradas sobre cada término

(Que existan referencias dedicadas a un término es también indicio de terminologicidad.)

precision

: Kohavi, R. (1996). Scaling up the precision of naive-bayes classifiers: A decision tree hybrid. Ponencia presentada en el Second International Conference on Knowledge Discovery and Data Mining, Portland, Oregon.