For MBR decoding, we instead leverage a similarity measure 8(6; 6’) to choose a translation using the model’s probability distribution P(e| f), which has support over a set of possible translations E. The Viterbi derivation 6* is the mode of this distribution.

Given any similarity measure 8 and a k-best list E, the minimum Bayes risk translation can be found by computing the similarity between all pairs of sentences in E, as in Algorithm 1.

An example of a linear similarity measure is bag-of-words precision, which can be written as:

The Bayes optimal decoding objective is to minimize risk based on the similarity measure used for evaluation.

Unfortunately, with a nonlinear similarity measure like BLEU, we must resort to approximating the expected loss using a k-best list, which accounts for only a tiny fraction of a model’s full posterior distribution.

We show that if the similarity measure is linear in features of a sentence, then computing expected similarity for all k sentences requires only k similarity evaluations.

For each term t in the dictionary and each SMS token 3,, we define a similarity measure a(t, 3,) that measures how closely the term 75 matches the SMS token 3,.

Combining the similarity measure and the inverse document frequency (idf) of t in the corpus, we define a weight function to (t, 3,).

The similarity measure and the weight function are discussed in detail in Section 5.1.

The weight function is a combination of similarity measure between t and Si and Inverse Document Frequency (idf) of t. The next two subsections explain the calculation of the similarity measure and the idf in detail.

5.1.1 Similarity Measure

For term t E D and token 3%- of the SMS, the similarity measure a(t, 81) between them is

Our research goal was to develop a directional similarity measure suitable for learning asymmetric relations, focusing empirically on lexical expansion.

This paper investigates the nature of directional (asymmetric) similarity measures , which aim to quantify distributional feature inclusion.

Then, word vectors are compared by some vector similarity measure .

This paper advocates the use of directional similarity measures for lexical expansion, and potentially for other tasks, based on distributional inclusion of feature vectors.

We tested our similarity measure by evaluating its utility for lexical expansion, compared with baselines of the LIN, WeedsPrec and balPrec measures

Next, for each similarity measure , the terms found similar to any of the event’s seeds (‘u —> seed’) were taken as expansion terms.

Often, distributional similarity measures are used to identify expanding terms (e.g.

More generally, directional relations are abundant in NLP settings, making symmetric similarity measures less suitable for their identification.

Despite the need for directional similarity measures , their investigation counts, to the best of our knowledge, only few works (Weeds and Weir, 2003; Geffet and Dagan, 2005; Bhagat et al., 2007; Szpektor and Dagan, 2008; Michelbacher et al., 2007) and is utterly lacking.

The best results are obtained with a novel second-order distributional similarity measure , and the positive effect is specially relevant for out-of-domain data.

We have empirically shown how automatically generated selectional preferences, using WordNet and distributional similarity measures , are able to effectively generalize lexical features and, thus, improve classification performance in a large-scale argument classification task on the CoNLL-2005 dataset.

Pantel and Lin (2000) obtained very good results using the distributional similarity measure defined by Lin (1998).

The second-order distributional similarity measures perform best overall, both in precision and recall.

We will refer to this similarity measure as simg‘n.

We will refer to these similarity measures as simE-ZE and simi’gg hereinafter.

Our method was evaluated for each (P1, P2, P3) combination and similarity measures J0 and 197,, separately.

In this paper, we propose a novel unsupervised approach that compares the major senses of a MWE and its semantic head using distributional similarity measures to test the compositionality of the MWE.

Lee (1999) shows that J performs better than other symmetric similarity measures such as cosine, Jensen-Shannon divergence, etc.

Given the major uses of a MWE and its semantic head, the MWE is considered as compositional, when the corresponding distributional similarity measure (Jc or 197,) value is above a parameter threshold, sim.

The best performing distributional similarity measure is an.

Siml and Sim2 respectively mean Formula 3.1 and Formula 3.2 are used in postprocessing as the similarity measure between

Sim2 achieves more performance improvement than Siml, which demonstrates the effectiveness of the similarity measure in Formula 3.2.

One simple and straightforward similarity measure is the J accard