Abstract | We propose a novel two-level model, which computes similarities between word-level vectors that are biased by topic-level context representations. |
Abstract | Evaluations on a naturally-distributed dataset show that our model significantly outperforms prior word-level and topic-level models. |
Background and Model Setting | However, while DIRT computes sim(v, 21’) over vectors in the original word-level space, topic-level models compute sim(d, d’, w) by measuring similarity of vectors in a reduced-dimensionality latent space. |
Background and Model Setting | slots in the original word-level space while biasing the similarity measure through topic-level context models. |
Introduction | To address this hypothesized caveat of prior context-sensitive rule scoring methods, we propose a novel generic scheme that integrates word-level and topic-level representations. |
Introduction | Rather than computing a single context-insensitive rule score, we compute a distinct word-level similarity score for each topic in an LDA model. |
Results | Specifically, topics are leveraged for high-level domain disambiguation, while fine grained word-level distributional similarity is computed for each rule under each such domain. |
Results | This result more explicitly shows the advantages of integrating word-level and context-sensitive topic-level similarities for differentiating valid and invalid contexts for rule applications. |
Two-level Context-sensitive Inference | Thus, our model computes similarity over word-level (rather than topic-level) argument vectors, while biasing it according to the specific argument words in the given rule application context. |
Two-level Context-sensitive Inference | The core of our contribution is thus defining the context-sensitive word-level vector similarity measure sim(v, v’ , w), as described in the remainder of this section. |
Two-level Context-sensitive Inference | This way, rather than replacing altogether the word-level values v(w) by the topic probabilities p(t|dv, w), as done in the topic-level models, we use the latter to only bias the former while preserving fine-grained word-level representations. |
Evaluation | Degorski (2011) uses concatenation of word-level base forms assigned by the tagger as a baseline. |
Introduction | According to the lemmatisation principles accompanying the NCP tagset, adjectives are lemmatised as masculine forms (gléwny), hence it is not sufficient to take word-level lemma nor the orthographic form to obtain phrase lemmatisation. |
Introduction | It is worth stressing that even the task of word-level lemmatisation is nontrivial for inflectional languages due to a large number of inflected forms and even larger number of syncretisms. |
Phrase lemmatisation as a tagging problem | To show the real setting, this time we give full NCP tags and word-level lemmas assigned as a result of tagging. |
Phrase lemmatisation as a tagging problem | The notation cas=n om means that to obtain the desired form (e. g. gléwne) you need to find an entry in a morphological dictionary that bears the same word-level lemma as the inflected form (gféwny) and a tag that results from taking the tag of the inflected form (adj : sgzinst :n:pos) and setting the value of the tagset attribute cas (grammatical case) to the value nom (nominative). |
Phrase lemmatisation as a tagging problem | Our idea is simple: by expressing phrase lemmatisation in terms of word-level transformations we can reduce the task to tagging problem and apply well known Machine Learning techniques that have been devised for solving such problems (e. g. CRF). |
Preparation of training data | The development set was enhanced with word-level transformations that were induced automatically in the following manner. |
Preparation of training data | The dictionary is stored as a set of (orthographic form, word-level lemma, tag). |
Preparation of training data | The task is to find a suitable transformation for the given inflected form from the original phrase, its tag and word-level lemma, but also given the desired form being part of human-assigned lemma. |
Evaluation | Segmentation performance is measured using word-level precision (P), recall (R), and F-measure (F). |
Evaluation | The best performance result achieved by G2 in our experiment is 81.7 in word-level F-measure, although this was obtained from search setting (c), using a heuristic p value 0.37. |
Evaluation | It would be interesting to confirm this by studying the correlation between description length and word-level F-measure. |