acl acl2013 acl2013-88 knowledge-graph by maker-knowledge-mining
Source: pdf
Author: Vlad Niculae ; Victoria Yaneva
Abstract: This paper presents work in progress towards automatic recognition and classification of comparisons and similes. Among possible applications, we discuss the place of this task in text simplification for readers with Autism Spectrum Disorders (ASD), who are known to have deficits in comprehending figurative language. We propose an approach to comparison recognition through the use of syntactic patterns. Keeping in mind the requirements of autistic readers, we discuss the properties relevant for distinguishing semantic criteria like figurativeness and abstractness.
Reference: text
sentIndex sentText sentNum sentScore
1 Computational considerations of comparisons and similes Vlad Niculae University of Wolverhampton vlad@vene . [sent-1, score-0.785]
2 ro Abstract This paper presents work in progress towards automatic recognition and classification of comparisons and similes. [sent-2, score-0.146]
3 Among possible applications, we discuss the place of this task in text simplification for readers with Autism Spectrum Disorders (ASD), who are known to have deficits in comprehending figurative language. [sent-3, score-0.356]
4 Keeping in mind the requirements of autistic readers, we discuss the properties relevant for distinguishing semantic criteria like figurativeness and abstractness. [sent-5, score-0.251]
5 The most obvious pattern, be like , illustrated by the following is example, but many subtler ways of building comparisons exist: “He was like his father, except he had a crooked nose and his ears were a little lopsided. [sent-8, score-0.284]
6 The simile is a figure of speech that builds on a comparison in order to exploit certain attributes of an entity in a striking manner. [sent-10, score-0.417]
7 According to the Oxford English Dictionary, what sets a simile apart from a comparison is that it compares “one thing with another thing of a different kind”1 . [sent-11, score-0.47]
8 a figure of speech involving the comparison of one thing with another thing of a different kind, used to make a description more emphatic or vivid (e. [sent-13, score-0.114]
9 Cratchit entered: flushed, but smiling proudly: with the pudding, like a speckled cannon-ball, so hard and firm, (. [sent-25, score-0.058]
10 Intuitively, the OED definition is confirmed by these two examples: a Christmas pudding and a cannon-ball are things of different kinds, whereas he and his father are things of the same kind (namely, human males). [sent-29, score-0.193]
11 As we shall see, the borderline which divides some similes and fixed expressions is the degree of conventionality. [sent-30, score-0.672]
12 In these cases, however, the link between the two entities is a pattern repeated so many times that it has consequently lost its innovativeness and turned into a dead metaphor (“as dead as a doornail”) or a conventional simile (sections 4. [sent-32, score-0.787]
13 The scholarly discussion of the simile has been controversial, especially with respect to its relative, the metaphor. [sent-36, score-0.356]
14 In other words, a metaphor links features of objects or events from two different, often incompatible domains, thus being a “realization of a crossdomain conceptual mapping” (Deignan, 2005). [sent-42, score-0.222]
15 We are interested in the parallel between similes and metaphors insofar as it points to an overlap. [sent-43, score-0.796]
16 There are types of similes that can be transformed into equivalent metaphors, and certain metaphors can be rewritten as similes, but neither set is included in the other. [sent-44, score-0.817]
17 (2004) suggest: some metaphors express things that cannot be expressed by similes, and vice versa. [sent-46, score-0.214]
18 In computational linguistics, similes have been neglected in favour of metaphor even more than in linguistics3 , despite the fact that comparisons have a structure that makes them rather amenable to automated processing. [sent-47, score-1.007]
19 In sections 2 we discuss one motivation for studying comparisons and similes: their simplification to language better suited for people with ASD. [sent-48, score-0.272]
20 Section 3 reviews related work on figurative language in NLP. [sent-49, score-0.23]
21 In section 4 we present the structure of comparisons and some associated patterns, emphasising the difficulties posed by the flexibility of language. [sent-50, score-0.146]
22 1 Autism and figurative language Highly abstract or figurative metaphors and similes may be problematic for certain groups of language users amongst which are people with different types of acquired language disorders (aphasias) or developmental ones like ASD. [sent-54, score-1.435]
23 com/ 3A Google Scholar search for papers containing the word linguistic have the word metaphor in the title approximately 5000 times, but simile only around 645 times. [sent-61, score-0.578]
24 In the ACL anthology, metaphor occurs around 1070 times while simile occurs 52 times. [sent-62, score-0.578]
25 People with autism, especially if they are children, experience disturbing confusion when confronted with figurative language. [sent-64, score-0.23]
26 The growing demand to overcome this barrier has led to the investigation of possible ways in which NLP can detect and simplify non-literal expressions in a text. [sent-71, score-0.055]
27 2 Comprehending similes People with ASD4 show almost no impairment in comprehending those similes which have literal meaning (Happ ´e, 1995). [sent-73, score-1.405]
28 This relative ease in processing is probably due to the fact that similes contain explicit markers (e. [sent-74, score-0.639]
29 like and as), which evoke comparison between two things in a certain aspect. [sent-76, score-0.176]
30 With regard to understanding figurative similes, Hobson (2012) describes in the case of fifteenyear-old L. [sent-77, score-0.258]
31 For example, like could be a verb, a noun, or a preposition, depending on the context. [sent-80, score-0.058]
32 Given that autistic people have problems understanding context (Skoyles, 2011), how would an autistic reader perceive the role of like in a more elaborate and ambiguous comparison? [sent-81, score-0.383]
33 Another possible linguistic reason for the impaired understanding of similes might be that like is used 4With level of cognitive ability corresponding to at least first level of Theory of Mind (Baron-Cohen et al. [sent-82, score-0.725]
34 , 1985) 90 ambiguously in many expressions which are neither similes nor comparisons, such as I feel like an ice cream or I feel like something is wrong. [sent-83, score-0.916]
35 Even if the expression does not include such an ambiguous use of like, there are other cases in which a person with autism might be misled. [sent-84, score-0.201]
36 For example, if the simile is highly figurative or abstract, it may be completely incomprehensible for people with ASD (e. [sent-85, score-0.627]
37 A step forward towards the simplification of such expressions is their identification and filtering of the ones that are not problematic. [sent-88, score-0.118]
38 3 Relevant literature Comprehensive theoretical investigations into the expressive power of similes can be found in (Bethlehem, 1996) and (Israel et al. [sent-90, score-0.639]
39 Weiner (1984) applies ontologies to discriminate simple literal and figurative comparisons (loosely using the term metaphor to refer to what we call the intersection of similes and metaphors). [sent-92, score-1.29]
40 Most of the recent computational linguistics research involving similes comes from Veale. [sent-93, score-0.639]
41 The Metaphor Mag- net system presented in (Veale and Li, 2012) supports queries against a rich ontology of metaphorical meanings and affects using the same simple simile patterns. [sent-97, score-0.383]
42 The Jigsaw Bard (Veale and Hao, 2011) is a thesaurus driven by figurative conventional similes extracted from the Google Ngram corpus. [sent-98, score-0.949]
43 The role played by figurative language in the field of text simplification has not been extensively studied outside of a few recent publications (Temnikova, 2012; Sˇtajner et al. [sent-99, score-0.315]
44 This distinction can be applied to similes: as slow as a snail is a conventional simile that evokes strong asso- ciation between slowness and snails. [sent-103, score-0.436]
45 Though figures of speech are good ways to exploit norms, figurative language can become conventional, and an exploitation can be literal (e. [sent-106, score-0.305]
46 2 Syntactic structure The breadth of comparisons and similes hasn’t been extensively studied, so there is no surprise in the small amount of coverage in computational linguistics research on the subject. [sent-114, score-0.785]
47 In order to develop a solid foundation for working with complex comparisons, we will follow and argue for the terminology from (Hanks, 2012), where the structure of a simile is analysed. [sent-115, score-0.356]
48 The same structure applies to comparisons, since as we have said, all similes are comparisons and they are indistinguishable syntactically. [sent-116, score-0.81]
49 C: the comparator: commonly a preposition (like or part aoraf an adjectival phrase (better than), it is the trigger word or phrase that marks the presence of a comparison. [sent-120, score-0.068]
50 Matches he eats like a pig and it is seen as a release. [sent-122, score-0.058]
51 An example (adapted from the BNC) of a simile involving all of the above would be: [He T] [looked E] [like C] [a broiled frog V], [hunched P] over his desk, grinning and satisfied. [sent-127, score-0.356]
52 Fishelov (1993) attributes this reordering to poetic simile, along with other deviations from the norm that he defines as non-poetic simile. [sent-129, score-0.074]
53 Fishelov even suggested that poetic similes can be found outside of poetic text, and vice versa. [sent-131, score-0.787]
54 We will therefore focus on exploitations that change the meaning. [sent-132, score-0.066]
55 More often than not, the property is left for the reader to deduce: [His mouth T] [tasted E] [like C] [the bottom of a parrot’s cage V] But even when all elements appear, the comparison may be ambiguous, as lexical choice in P and in E lead to various degrees of specificity. [sent-133, score-0.165]
56 For example replacing the word tasted, which forms the E in the example above, with the more general predicator is, results in a simile that might have the same meaning, but is more difficult to decode. [sent-134, score-0.389]
57 On the other hand, the whole V phrase the bottom of a parrot’s cage, which is an euphemistic metonymy, could be substituted with its concrete, literal meaning thus transforming the creative simile into what might be a conventional pattern. [sent-135, score-0.489]
58 Nested figures of speech can also occur at this level, for example the insertion of a metaphorical and synesthetic P: it tasted [dirty P], like a parrot’s cage. [sent-136, score-0.159]
59 We consider the eventuality E as the syntactic core of the comparison structure. [sent-137, score-0.077]
60 Roncero (2006) pointed out that for certain common similes (e. [sent-141, score-0.66]
61 love is like a rose) found on the Internet, it is likely that an explanation of the shared property follows, whereas for all topicvehicle pairs studied, the corresponding metaphor is less often explained. [sent-143, score-0.304]
62 However, these simpler similes form a special case, as most similes cannot be made into metaphors (Hanks, 2012). [sent-144, score-1.435]
63 3 Comparisons without like Hanks (2012) observes that there are plenty of other ways to make a simile in addition to using like or as. [sent-146, score-0.494]
64 Most definitions of similes indeed claim that there are more possible comparators, but examples are elusive. [sent-147, score-0.639]
65 The task of understanding similes may be hard to achieve. [sent-155, score-0.667]
66 Often similes claim that a property is present or absent, but this is not always the case. [sent-159, score-0.663]
67 2 Dataset At the moment there is no available dataset for comparison and simile recognition and classification. [sent-161, score-0.396]
68 We have begun our investigation and developed the patterns on a toy dataset consisting of the examples from (Hanks, 2005), which are comparisons, similes and other ambiguous uses of the preposition like extracted from the BNC. [sent-162, score-0.816]
69 We also evaluated the system on around 500 sentences containing like and as from the BNC and the VUAMC5. [sent-163, score-0.058]
70 1 Comparison pattern matching We have seen that similes are a subset of comparisons and follow comparison structures. [sent-172, score-0.864]
71 It enhances the constituency-based parse tree with additional roles and arguments by applying rules and resources like Propbank. [sent-176, score-0.058]
72 The like and as comparators form the GLARF-style patterns shown in figure 1. [sent-177, score-0.148]
73 If the subtree rooted under it matches certain filters, then we assign to the root the role of E and the arguments can fill the other slots. [sent-179, score-0.075]
74 Expressions like in hold your hands like this are mistaken as comparisons. [sent-186, score-0.141]
75 big earners like doctors and airline pilots but incorrectly matches semantically ambiguous uses of feel like. [sent-189, score-0.184]
76 Most errors are therefore returning spurious matches, as opposed to like, where most errors are omissions This suggests that each trigger word behaves differently, and therefore robustness across patterns is important. [sent-191, score-0.089]
77 Overall, our method handles typical comparisons in short sentences rather well. [sent-192, score-0.146]
78 2 Discovering new patterns Using a seed-based semi-supervised iterative process, we plan to identify most of the frequent structures used to build conventional comparisons. [sent-197, score-0.133]
79 33 (a) Counts of 40 examples with like from the (b) Proportions of 410 development set in (Hanks, 2005). [sent-204, score-0.058]
80 Partial match examples with like from BNC P = 94%, R = 88%. [sent-205, score-0.08]
81 other with the like pattern will occur in other tactical patterns or lexical collocations. [sent-220, score-0.15]
82 1 Classifying comparisons The phrases that match patterns like the ones described are not necessarily comparisons. [sent-224, score-0.279]
83 Due to ambiguities, sentences such as I feel like an ice cream are indistinguishable from comparisons in our model. [sent-225, score-0.319]
84 Another aspect we would like to distinguish is whether an instance of a pattern is a simile or not. [sent-226, score-0.453]
85 Semantic features from an ontology like the one used in PDEV6, or a more comprehensive work such as WordNet7, can carry the information whether T and V belong to similar semantic categories. [sent-228, score-0.058]
86 2 Conventional similes It may also be of interest to decide whether an instance is conventional or creative. [sent-232, score-0.719]
87 3 Simplification The goal of text simplification is to generate syntactically well-formed that is easier to language9 6http : / / deb . [sent-237, score-0.085]
88 edu / 8Care must be taken to avoid contradictions from exploitations: The aircraft is like a rock or is built like a rock seems like a conventional simile, but The aircraft would gently skip like a rock and then settle down on the surface of the ocean (Example from the BNC) is unconventional. [sent-242, score-0.507]
89 We can think of his mouth tasted like the bottom of a parrot’s cage as a way to express taste(his mouth; very bad). [sent-246, score-0.233]
90 Useful resources are corpus occurrence counts of related phrases, word similarity and relatedness, and conventional associations. [sent-249, score-0.08]
91 6 Conclusions and future work The problem of automatic identification of similes has its place in the paradigm of text simplification for people with language impairments. [sent-250, score-0.765]
92 In particular, people with ASD have difficulties understanding figurative language. [sent-251, score-0.299]
93 We applied the idea of comparison patterns to match subtrees of an enhanced parse tree to easily match comparison structures and their constituents. [sent-252, score-0.177]
94 This lead us to investigate corpus-driven mining of new comparison patterns, to go beyond like and as. [sent-253, score-0.098]
95 We are working on semi-automatically developing a dataset of comparisons and ambiguous noncomparisons, labelled with the interesting properties and with a focus on pattern variety and am- biguous cases. [sent-254, score-0.219]
96 We plan to perform extrinsic evaluation with respect to tasks like text simplification, textual entailment and machine translation. [sent-256, score-0.058]
97 Understanding minds and metaphors: Insights from the study of figurative language in autism. [sent-299, score-0.23]
98 Autism, literal language and concrete thinking: Some developmental considerations. [sent-304, score-0.079]
99 A comparative study of figurative language in children with autistic spectrum disorders. [sent-322, score-0.364]
100 The atypical development of metaphor and metonymy comprehension in children with autism. [sent-336, score-0.32]
wordName wordTfidf (topN-words)
[('similes', 0.639), ('simile', 0.356), ('figurative', 0.23), ('metaphor', 0.222), ('autism', 0.167), ('metaphors', 0.157), ('comparisons', 0.146), ('autistic', 0.111), ('christmas', 0.098), ('asd', 0.093), ('veale', 0.085), ('simplification', 0.085), ('hanks', 0.08), ('conventional', 0.08), ('happ', 0.074), ('poetic', 0.074), ('tasted', 0.074), ('wolverhampton', 0.074), ('bnc', 0.07), ('bethlehem', 0.066), ('exploitations', 0.066), ('parrot', 0.066), ('like', 0.058), ('norms', 0.057), ('things', 0.057), ('cage', 0.056), ('dickens', 0.056), ('figurativeness', 0.056), ('fishelov', 0.056), ('pudding', 0.056), ('matches', 0.054), ('patterns', 0.053), ('literal', 0.053), ('mouth', 0.045), ('atypical', 0.045), ('dead', 0.045), ('oed', 0.045), ('rock', 0.045), ('israel', 0.044), ('people', 0.041), ('comprehending', 0.041), ('comparison', 0.04), ('comparator', 0.039), ('pattern', 0.039), ('feel', 0.038), ('vehicle', 0.037), ('aristoteles', 0.037), ('comparators', 0.037), ('conventionality', 0.037), ('defiguration', 0.037), ('deignan', 0.037), ('doornail', 0.037), ('eventuality', 0.037), ('jigsaw', 0.037), ('rhetoric', 0.037), ('roncero', 0.037), ('rundblad', 0.037), ('skoyles', 0.037), ('steen', 0.037), ('tajner', 0.037), ('thing', 0.037), ('trigger', 0.036), ('ambiguous', 0.034), ('expressions', 0.033), ('predicator', 0.033), ('disorders', 0.033), ('firm', 0.033), ('impairment', 0.033), ('vlad', 0.033), ('yanfen', 0.033), ('preposition', 0.032), ('tony', 0.032), ('metonymy', 0.03), ('aircraft', 0.03), ('mackay', 0.03), ('poetics', 0.03), ('converging', 0.028), ('understanding', 0.028), ('lakoff', 0.027), ('cream', 0.027), ('metaphorical', 0.027), ('carol', 0.027), ('oxford', 0.026), ('conventionally', 0.026), ('developmental', 0.026), ('mind', 0.026), ('indistinguishable', 0.025), ('february', 0.025), ('ice', 0.025), ('hands', 0.025), ('property', 0.024), ('father', 0.023), ('comprehension', 0.023), ('spectrum', 0.023), ('ways', 0.022), ('match', 0.022), ('communication', 0.022), ('simplifying', 0.022), ('proportions', 0.022), ('certain', 0.021)]
simIndex simValue paperId paperTitle
same-paper 1 1.0 88 acl-2013-Computational considerations of comparisons and similes
Author: Vlad Niculae ; Victoria Yaneva
Abstract: This paper presents work in progress towards automatic recognition and classification of comparisons and similes. Among possible applications, we discuss the place of this task in text simplification for readers with Autism Spectrum Disorders (ASD), who are known to have deficits in comprehending figurative language. We propose an approach to comparison recognition through the use of syntactic patterns. Keeping in mind the requirements of autistic readers, we discuss the properties relevant for distinguishing semantic criteria like figurativeness and abstractness.
2 0.24236143 116 acl-2013-Detecting Metaphor by Contextual Analogy
Author: Eirini Florou
Abstract: As one of the most challenging issues in NLP, metaphor identification and its interpretation have seen many models and methods proposed. This paper presents a study on metaphor identification based on the semantic similarity between literal and non literal meanings of words that can appear at the same context.
3 0.16769619 253 acl-2013-Multilingual Affect Polarity and Valence Prediction in Metaphor-Rich Texts
Author: Zornitsa Kozareva
Abstract: Metaphor is an important way of conveying the affect of people, hence understanding how people use metaphors to convey affect is important for the communication between individuals and increases cohesion if the perceived affect of the concrete example is the same for the two individuals. Therefore, building computational models that can automatically identify the affect in metaphor-rich texts like “The team captain is a rock.”, “Time is money.”, “My lawyer is a shark.” is an important challenging problem, which has been of great interest to the research community. To solve this task, we have collected and manually annotated the affect of metaphor-rich texts for four languages. We present novel algorithms that integrate triggers for cognitive, affective, perceptual and social processes with stylistic and lexical information. By running evaluations on datasets in English, Spanish, Russian and Farsi, we show that the developed affect polarity and valence prediction technology of metaphor-rich texts is portable and works equally well for different languages.
4 0.13670526 96 acl-2013-Creating Similarity: Lateral Thinking for Vertical Similarity Judgments
Author: Tony Veale ; Guofu Li
Abstract: Just as observing is more than just seeing, comparing is far more than mere matching. It takes understanding, and even inventiveness, to discern a useful basis for judging two ideas as similar in a particular context, especially when our perspective is shaped by an act of linguistic creativity such as metaphor, simile or analogy. Structured resources such as WordNet offer a convenient hierarchical means for converging on a common ground for comparison, but offer little support for the divergent thinking that is needed to creatively view one concept as another. We describe such a means here, by showing how the web can be used to harvest many divergent views for many familiar ideas. These lateral views complement the vertical views of WordNet, and support a system for idea exploration called Thesaurus Rex. We show also how Thesaurus Rex supports a novel, generative similarity measure for WordNet. 1 Seeing is Believing (and Creating) Similarity is a cognitive phenomenon that is both complex and subjective, yet for practical reasons it is often modeled as if it were simple and objective. This makes sense for the many situations where we want to align our similarity judgments with those of others, and thus focus on the same conventional properties that others are also likely to focus upon. This reliance on the consensus viewpoint explains why WordNet (Fellbaum, 1998) has proven so useful as a basis for computational measures of lexico-semantic similarity Guofu Li School of Computer Science and Informatics, University College Dublin, Belfield, Dublin D2, Ireland. l .guo fu . l gmai l i @ .com (e.g. see Pederson et al. 2004, Budanitsky & Hirst, 2006; Seco et al. 2006). These measures reduce the similarity of two lexical concepts to a single number, by viewing similarity as an objective estimate of the overlap in their salient qualities. This convenient perspective is poorly suited to creative or insightful comparisons, but it is sufficient for the many mundane comparisons we often perform in daily life, such as when we organize books or look for items in a supermarket. So if we do not know in which aisle to locate a given item (such as oatmeal), we may tacitly know how to locate a similar product (such as cornflakes) and orient ourselves accordingly. Yet there are occasions when the recognition of similarities spurs the creation of similarities, when the act of comparison spurs us to invent new ways of looking at an idea. By placing pop tarts in the breakfast aisle, food manufacturers encourage us to view them as a breakfast food that is not dissimilar to oatmeal or cornflakes. When ex-PM Tony Blair published his memoirs, a mischievous activist encouraged others to move his book from Biography to Fiction in bookshops, in the hope that buyers would see it in a new light. Whenever we use a novel metaphor to convey a non-obvious viewpoint on a topic, such as “cigarettes are time bombs”, the comparison may spur us to insight, to see aspects of the topic that make it more similar to the vehicle (see Ortony, 1979; Veale & Hao, 2007). In formal terms, assume agent A has an insight about concept X, and uses the metaphor X is a Y to also provoke this insight in agent B. To arrive at this insight for itself, B must intuit what X and Y have in common. But this commonality is surely more than a standard categorization of X, or else it would not count as an insight about X. To understand the metaphor, B must place X 660 Proce dingSsof oifa, th Beu 5l1gsarti Aan,An u aglu Mste 4e-ti9n2g 0 o1f3 t.he ?c A2s0s1o3ci Aatsiosonc fioartio Cno fmorpu Ctoamtiopnuatalt Lioin gauli Lsitnicgsu,i psatgices 6 0–670, in a new category, so that X can be seen as more similar to Y. Metaphors shape the way we per- ceive the world by re-shaping the way we make similarity judgments. So if we want to imbue computers with the ability to make and to understand creative metaphors, we must first give them the ability to look beyond the narrow viewpoints of conventional resources. Any measure that models similarity as an objective function of a conventional worldview employs a convergent thought process. Using WordNet, for instance, a similarity measure can vertically converge on a common superordinate category of both inputs, and generate a single numeric result based on their distance to, and the information content of, this common generalization. So to find the most conventional ways of seeing a lexical concept, one simply ascends a narrowing concept hierarchy, using a process de Bono (1970) calls vertical thinking. To find novel, non-obvious and useful ways of looking at a lexical concept, one must use what Guilford (1967) calls divergent thinking and what de Bono calls lateral thinking. These processes cut across familiar category boundaries, to simultaneously place a concept in many different categories so that we can see it in many different ways. de Bono argues that vertical thinking is selective while lateral thinking is generative. Whereas vertical thinking concerns itself with the “right” way or a single “best” way of looking at things, lateral thinking focuses on producing alternatives to the status quo. To be as useful for creative tasks as they are for conventional tasks, we need to re-imagine our computational similarity measures as generative rather than selective, expansive rather than reductive, divergent as well as convergent and lateral as well as vertical. Though WordNet is ideally structured to support vertical, convergent reasoning, its comprehensive nature means it can also be used as a solid foundation for building a more lateral and divergent model of similarity. Here we will use the web as a source of diverse perspectives on familiar ideas, to complement the conventional and often narrow views codified by WordNet. Section 2 provides a brief overview of past work in the area of similarity measurement, before section 3 describes a simple bootstrapping loop for acquiring richly diverse perspectives from the web for a wide variety of familiar ideas. These perspectives are used to enhance a Word- Net-based measure of lexico-semantic similarity in section 4, by broadening the range of informative viewpoints the measure can select from. Similarity is thus modeled as a process that is both generative and selective. This lateral-andvertical approach is evaluated in section 5, on the Miller & Charles (1991) data-set. A web app for the lateral exploration of diverse viewpoints, named Thesaurus Rex, is also presented, before closing remarks are offered in section 6. 2 Related Work and Ideas WordNet’s taxonomic organization of nounsenses and verb-senses – in which very general categories are successively divided into increasingly informative sub-categories or instancelevel ideas – allows us to gauge the overlap in information content, and thus of meaning, of two lexical concepts. We need only identify the deepest point in the taxonomy at which this content starts to diverge. This point of divergence is often called the LCS, or least common subsumer, of two concepts (Pederson et al., 2004). Since sub-categories add new properties to those they inherit from their parents – Aristotle called these properties the differentia that stop a category system from trivially collapsing into itself – the depth of a lexical concept in a taxonomy is an intuitive proxy for its information content. Wu & Palmer (1994) use the depth of a lexical concept in the WordNet hierarchy as such a proxy, and thereby estimate the similarity of two lexical concepts as twice the depth of their LCS divided by the sum of their individual depths. Leacock and Chodorow (1998) instead use the length of the shortest path between two concepts as a proxy for the conceptual distance between them. To connect any two ideas in a hierarchical system, one must vertically ascend the hierarchy from one concept, change direction at a potential LCS, and then descend the hierarchy to reach the second concept. (Aristotle was also first to suggest this approach in his Poetics). Leacock and Chodorow normalize the length of this path by dividing its size (in nodes) by twice the depth of the deepest concept in the hierarchy; the latter is an upper bound on the distance between any two concepts in the hierarchy. Negating the log of this normalized length yields a corresponding similarity score. While the role of an LCS is merely implied in Leacock and Chodorow’s use of a shortest path, the LCS is pivotal nonetheless, and like that of Wu & Palmer, the approach uses an essentially vertical reasoning process to identify a single “best” generalization. Depth is a convenient proxy for information content, but more nuanced proxies can yield 661 more rounded similarity measures. Resnick (1995) draws on information theory to define the information content of a lexical concept as the negative log likelihood of its occurrence in a corpus, either explicitly (via a direct mention) or by presupposition (via a mention of any of its sub-categories or instances). Since the likelihood of a general category occurring in a corpus is higher than that of any of its sub-categories or instances, such categories are more predictable, and less informative, than rarer categories whose occurrences are less predictable and thus more informative. The negative log likelihood of the most informative LCS of two lexical concepts offers a reliable estimate of the amount of infor- mation shared by those concepts, and thus a good estimate of their similarity. Lin (1998) combines the intuitions behind Resnick’s metric and that of Wu and Palmer to estimate the similarity of two lexical concepts as an information ratio: twice the information content of their LCS divided by the sum of their individual information contents. Jiang and Conrath (1997) consider the converse notion of dissimilarity, noting that two lexical concepts are dissimilar to the extent that each contains information that is not shared by the other. So if the information content of their most informative LCS is a good measure of what they do share, then the sum of their individual information contents, minus twice the content of their most informative LCS, is a reliable estimate of their dissimilarity. Seco et al. (2006) presents a minor innovation, showing how Resnick’s notion of information content can be calculated without the use of an external corpus. Rather, when using Resnick’s metric (or that of Lin, or Jiang and Conrath) for measuring the similarity of lexical concepts in WordNet, one can use the category structure of WordNet itself to estimate infor- mation content. Typically, the more general a concept, the more descendants it will possess. Seco et al. thus estimate the information content of a lexical concept as the log of the sum of all its unique descendants (both direct and indirect), divided by the log of the total number of concepts in the entire hierarchy. Not only is this intrinsic view of information content convenient to use, without recourse to an external corpus, Seco et al. show that it offers a better estimate of information content than its extrinsic, corpus-based alternatives, as measured relative to average human similarity ratings for the 30 word-pairs in the Miller & Charles (1991) test set. A similarity measure can draw on other sources of information besides WordNet’s category structures. One might eke out additional information from WordNet’s textual glosses, as in Lesk (1986), or use category structures other than those offered by WordNet. Looking beyond WordNet, entries in the online encyclopedia Wikipedia are not only connected by a dense topology of lateral links, they are also organized by a rich hierarchy of overlapping categories. Strube and Ponzetto (2006) show how Wikipedia can support a measure of similarity (and relatedness) that better approximates human judgments than many WordNet-based measures. Nonetheless, WordNet can be a valuable component of a hybrid measure, and Agirre et al. (2009) use an SVM (support vector machine) to combine information from WordNet with information harvested from the web. Their best similarity measure achieves a remarkable 0.93 correlation with human judgments on the Miller & Charles word-pair set. Similarity is not always applied to pairs of concepts; it is sometimes analogically applied to pairs of pairs of concepts, as in proportional analogies of the form A is to B as C is to D (e.g., hacks are to writers as mercenaries are to soldiers, or chisels are to sculptors as scalpels are to surgeons). In such analogies, one is really assessing the similarity of the unstated relationship between each pair of concepts: thus, mercenaries are soldiers whose allegiance is paid for, much as hacks are writers with income-driven loyalties; sculptors use chisels to carve stone, while surgeons use scalpels to cut or carve flesh. Veale (2004) used WordNet to assess the similarity of A:B to C:D as a function of the combined similarity of A to C and of B to D. In contrast, Turney (2005) used the web to pursue a more divergent course, to represent the tacit relationships of A to B and of C to D as points in a highdimensional space. The dimensions of this space initially correspond to linking phrases on the web, before these dimensions are significantly reduced using singular value decomposition. In the infamous SAT test, an analogy A:B::C:D has four other pairs of concepts that serve as likely distractors (e.g. singer:songwriter for hack:writer) and the goal is to choose the most appropriate C:D pair for a given A:B pairing. Using variants of Wu and Palmer (1994) on the 374 SAT analogies of Turney (2005), Veale (2004) reports a success rate of 38–44% using only WordNet-based similarity. In contrast, Turney (2005) reports up to 55% success on the same analogies, partly because his approach aims 662 to match implicit relations rather than explicit concepts, and in part because it uses a divergent process to gather from the web as rich a perspec- tive as it can on these latent relationships. 2.1 Clever Comparisons Create Similarity Each of these approaches to similarity is a user of information, rather than a creator, and each fails to capture how a creative comparison (such as a metaphor) can spur a listener to view a topic from an atypical perspective. Camac & Glucksberg (1984) provide experimental evidence for the claim that “metaphors do not use preexisting associations to achieve their effects [… ] people use metaphors to create new relations between concepts.” They also offer a salutary reminder of an often overlooked fact: every comparison exploits information, but each is also a source of new information in its own right. Thus, “this cola is acid” reveals a different perspective on cola (e.g. as a corrosive substance or an irritating food) than “this acid is cola” highlights for acid (such as e.g., a familiar substance) Veale & Keane (1994) model the role of similarity in realizing the long-term perlocutionary effect of an informative comparison. For example, to compare surgeons to butchers is to encourage one to see all surgeons as more bloody, … crude or careless. The reverse comparison, of butchers to surgeons, encourages one to see butchers as more skilled and precise. Veale & Keane present a network model of memory, called Sapper, in which activation can spread between related concepts, thus allowing one concept to prime the properties of a neighbor. To interpret an analogy, Sapper lays down new activation-carrying bridges in memory between analogical counterparts, such as between surgeon & butcher, flesh & meat, and scalpel & cleaver. Comparisons can thus have lasting effects on how Sapper sees the world, changing the pattern of activation that arises when it primes a concept. Veale (2003) adopts a similarly dynamic view of similarity in WordNet, showing how an analogical comparison can result in the automatic addition of new categories and relations to WordNet itself. Veale considers the problem of finding an analogical mapping between different parts of WordNet’s noun-sense hierarchy, such as between instances of Greek god and Norse god, or between the letters of different alphabets, such as of Greek and Hebrew. But no structural similarity measure for WordNet exhibits enough discernment to e.g. assign a higher similarity to Zeus & Odin (each is the supreme deity of its pantheon) than to a pairing of Zeus and any other Norse god, just as no structural measure will assign a higher similarity to Alpha & Aleph or to Beta & Beth than to any random letter pairing. A fine-grained category hierarchy permits fine-grained similarity judgments, and though WordNet is useful, its sense hierarchies are not especially fine-grained. However, we can automatically make WordNet subtler and more discerning, by adding new fine-grained categories to unite lexical concepts whose similarity is not reflected by any existing categories. Veale (2003) shows how a property that is found in the glosses of two lexical concepts, of the same depth, can be combined with their LCS to yield a new fine-grained parent category, so e.g. “supreme” + deity = Supreme-deity (for Odin, Zeus, Jupiter, etc.) and “1 st” + letter = 1st-letter (for Alpha, Aleph, etc.) Selected aspects of the textual similarity of two WordNet glosses – the key to similarity in Lesk (1986) – can thus be reified into an explicitly categorical WordNet form. 3 Divergent (Re)Categorization To tap into a richer source of concept properties than WordNet’s glosses, we can use web ngrams. Consider these descriptions of a cowboy from the Google n-grams (Brants & Franz, 2006). The numbers to the right are Google frequency counts. a lonesome cowboy 432 a mounted cowboy 122 a grizzled cowboy 74 a swaggering cowboy 68 To find the stable properties that can underpin a meaningful fine-grained category for cowboy, we must seek out the properties that are so often presupposed to be salient of all cowboys that one can use them to anchor a simile, such as
5 0.054298874 3 acl-2013-A Comparison of Techniques to Automatically Identify Complex Words.
Author: Matthew Shardlow
Abstract: Identifying complex words (CWs) is an important, yet often overlooked, task within lexical simplification (The process of automatically replacing CWs with simpler alternatives). If too many words are identified then substitutions may be made erroneously, leading to a loss of meaning. If too few words are identified then those which impede a user’s understanding may be missed, resulting in a complex final text. This paper addresses the task of evaluating different methods for CW identification. A corpus of sentences with annotated CWs is mined from Simple Wikipedia edit histories, which is then used as the basis for several experiments. Firstly, the corpus design is explained and the results of the validation experiments using human judges are reported. Experiments are carried out into the CW identification techniques of: simplifying everything, frequency thresholding and training a support vector machine. These are based upon previous approaches to the task and show that thresholding does not perform significantly differently to the more na¨ ıve technique of simplifying everything. The support vector machine achieves a slight increase in precision over the other two methods, but at the cost of a dramatic trade off in recall.
6 0.047823526 250 acl-2013-Models of Translation Competitions
7 0.047269005 322 acl-2013-Simple, readable sub-sentences
8 0.044247348 344 acl-2013-The Effects of Lexical Resource Quality on Preference Violation Detection
9 0.041853216 56 acl-2013-Argument Inference from Relevant Event Mentions in Chinese Argument Extraction
10 0.041156426 65 acl-2013-BRAINSUP: Brainstorming Support for Creative Sentence Generation
11 0.040935099 194 acl-2013-Improving Text Simplification Language Modeling Using Unsimplified Text Data
12 0.040238936 12 acl-2013-A New Set of Norms for Semantic Relatedness Measures
13 0.039273571 306 acl-2013-SPred: Large-scale Harvesting of Semantic Predicates
14 0.035949755 206 acl-2013-Joint Event Extraction via Structured Prediction with Global Features
15 0.03423252 249 acl-2013-Models of Semantic Representation with Visual Attributes
16 0.033810358 135 acl-2013-English-to-Russian MT evaluation campaign
17 0.03241355 238 acl-2013-Measuring semantic content in distributional vectors
18 0.032303847 160 acl-2013-Fine-grained Semantic Typing of Emerging Entities
19 0.031856555 373 acl-2013-Using Conceptual Class Attributes to Characterize Social Media Users
20 0.031088755 190 acl-2013-Implicatures and Nested Beliefs in Approximate Decentralized-POMDPs
topicId topicWeight
[(0, 0.102), (1, 0.041), (2, -0.003), (3, -0.056), (4, -0.052), (5, -0.029), (6, -0.037), (7, 0.042), (8, 0.029), (9, 0.024), (10, -0.038), (11, 0.025), (12, -0.057), (13, 0.004), (14, 0.0), (15, -0.022), (16, -0.004), (17, 0.019), (18, 0.03), (19, -0.007), (20, 0.029), (21, -0.043), (22, 0.042), (23, -0.041), (24, 0.071), (25, 0.011), (26, -0.103), (27, 0.012), (28, -0.043), (29, 0.065), (30, -0.001), (31, -0.223), (32, -0.008), (33, 0.002), (34, 0.115), (35, 0.092), (36, 0.207), (37, -0.076), (38, 0.102), (39, 0.02), (40, -0.07), (41, 0.096), (42, 0.025), (43, 0.139), (44, -0.053), (45, -0.075), (46, 0.004), (47, 0.001), (48, 0.075), (49, 0.122)]
simIndex simValue paperId paperTitle
same-paper 1 0.91916645 88 acl-2013-Computational considerations of comparisons and similes
Author: Vlad Niculae ; Victoria Yaneva
Abstract: This paper presents work in progress towards automatic recognition and classification of comparisons and similes. Among possible applications, we discuss the place of this task in text simplification for readers with Autism Spectrum Disorders (ASD), who are known to have deficits in comprehending figurative language. We propose an approach to comparison recognition through the use of syntactic patterns. Keeping in mind the requirements of autistic readers, we discuss the properties relevant for distinguishing semantic criteria like figurativeness and abstractness.
2 0.78493875 253 acl-2013-Multilingual Affect Polarity and Valence Prediction in Metaphor-Rich Texts
Author: Zornitsa Kozareva
Abstract: Metaphor is an important way of conveying the affect of people, hence understanding how people use metaphors to convey affect is important for the communication between individuals and increases cohesion if the perceived affect of the concrete example is the same for the two individuals. Therefore, building computational models that can automatically identify the affect in metaphor-rich texts like “The team captain is a rock.”, “Time is money.”, “My lawyer is a shark.” is an important challenging problem, which has been of great interest to the research community. To solve this task, we have collected and manually annotated the affect of metaphor-rich texts for four languages. We present novel algorithms that integrate triggers for cognitive, affective, perceptual and social processes with stylistic and lexical information. By running evaluations on datasets in English, Spanish, Russian and Farsi, we show that the developed affect polarity and valence prediction technology of metaphor-rich texts is portable and works equally well for different languages.
3 0.72662765 116 acl-2013-Detecting Metaphor by Contextual Analogy
Author: Eirini Florou
Abstract: As one of the most challenging issues in NLP, metaphor identification and its interpretation have seen many models and methods proposed. This paper presents a study on metaphor identification based on the semantic similarity between literal and non literal meanings of words that can appear at the same context.
4 0.53323978 96 acl-2013-Creating Similarity: Lateral Thinking for Vertical Similarity Judgments
Author: Tony Veale ; Guofu Li
Abstract: Just as observing is more than just seeing, comparing is far more than mere matching. It takes understanding, and even inventiveness, to discern a useful basis for judging two ideas as similar in a particular context, especially when our perspective is shaped by an act of linguistic creativity such as metaphor, simile or analogy. Structured resources such as WordNet offer a convenient hierarchical means for converging on a common ground for comparison, but offer little support for the divergent thinking that is needed to creatively view one concept as another. We describe such a means here, by showing how the web can be used to harvest many divergent views for many familiar ideas. These lateral views complement the vertical views of WordNet, and support a system for idea exploration called Thesaurus Rex. We show also how Thesaurus Rex supports a novel, generative similarity measure for WordNet. 1 Seeing is Believing (and Creating) Similarity is a cognitive phenomenon that is both complex and subjective, yet for practical reasons it is often modeled as if it were simple and objective. This makes sense for the many situations where we want to align our similarity judgments with those of others, and thus focus on the same conventional properties that others are also likely to focus upon. This reliance on the consensus viewpoint explains why WordNet (Fellbaum, 1998) has proven so useful as a basis for computational measures of lexico-semantic similarity Guofu Li School of Computer Science and Informatics, University College Dublin, Belfield, Dublin D2, Ireland. l .guo fu . l gmai l i @ .com (e.g. see Pederson et al. 2004, Budanitsky & Hirst, 2006; Seco et al. 2006). These measures reduce the similarity of two lexical concepts to a single number, by viewing similarity as an objective estimate of the overlap in their salient qualities. This convenient perspective is poorly suited to creative or insightful comparisons, but it is sufficient for the many mundane comparisons we often perform in daily life, such as when we organize books or look for items in a supermarket. So if we do not know in which aisle to locate a given item (such as oatmeal), we may tacitly know how to locate a similar product (such as cornflakes) and orient ourselves accordingly. Yet there are occasions when the recognition of similarities spurs the creation of similarities, when the act of comparison spurs us to invent new ways of looking at an idea. By placing pop tarts in the breakfast aisle, food manufacturers encourage us to view them as a breakfast food that is not dissimilar to oatmeal or cornflakes. When ex-PM Tony Blair published his memoirs, a mischievous activist encouraged others to move his book from Biography to Fiction in bookshops, in the hope that buyers would see it in a new light. Whenever we use a novel metaphor to convey a non-obvious viewpoint on a topic, such as “cigarettes are time bombs”, the comparison may spur us to insight, to see aspects of the topic that make it more similar to the vehicle (see Ortony, 1979; Veale & Hao, 2007). In formal terms, assume agent A has an insight about concept X, and uses the metaphor X is a Y to also provoke this insight in agent B. To arrive at this insight for itself, B must intuit what X and Y have in common. But this commonality is surely more than a standard categorization of X, or else it would not count as an insight about X. To understand the metaphor, B must place X 660 Proce dingSsof oifa, th Beu 5l1gsarti Aan,An u aglu Mste 4e-ti9n2g 0 o1f3 t.he ?c A2s0s1o3ci Aatsiosonc fioartio Cno fmorpu Ctoamtiopnuatalt Lioin gauli Lsitnicgsu,i psatgices 6 0–670, in a new category, so that X can be seen as more similar to Y. Metaphors shape the way we per- ceive the world by re-shaping the way we make similarity judgments. So if we want to imbue computers with the ability to make and to understand creative metaphors, we must first give them the ability to look beyond the narrow viewpoints of conventional resources. Any measure that models similarity as an objective function of a conventional worldview employs a convergent thought process. Using WordNet, for instance, a similarity measure can vertically converge on a common superordinate category of both inputs, and generate a single numeric result based on their distance to, and the information content of, this common generalization. So to find the most conventional ways of seeing a lexical concept, one simply ascends a narrowing concept hierarchy, using a process de Bono (1970) calls vertical thinking. To find novel, non-obvious and useful ways of looking at a lexical concept, one must use what Guilford (1967) calls divergent thinking and what de Bono calls lateral thinking. These processes cut across familiar category boundaries, to simultaneously place a concept in many different categories so that we can see it in many different ways. de Bono argues that vertical thinking is selective while lateral thinking is generative. Whereas vertical thinking concerns itself with the “right” way or a single “best” way of looking at things, lateral thinking focuses on producing alternatives to the status quo. To be as useful for creative tasks as they are for conventional tasks, we need to re-imagine our computational similarity measures as generative rather than selective, expansive rather than reductive, divergent as well as convergent and lateral as well as vertical. Though WordNet is ideally structured to support vertical, convergent reasoning, its comprehensive nature means it can also be used as a solid foundation for building a more lateral and divergent model of similarity. Here we will use the web as a source of diverse perspectives on familiar ideas, to complement the conventional and often narrow views codified by WordNet. Section 2 provides a brief overview of past work in the area of similarity measurement, before section 3 describes a simple bootstrapping loop for acquiring richly diverse perspectives from the web for a wide variety of familiar ideas. These perspectives are used to enhance a Word- Net-based measure of lexico-semantic similarity in section 4, by broadening the range of informative viewpoints the measure can select from. Similarity is thus modeled as a process that is both generative and selective. This lateral-andvertical approach is evaluated in section 5, on the Miller & Charles (1991) data-set. A web app for the lateral exploration of diverse viewpoints, named Thesaurus Rex, is also presented, before closing remarks are offered in section 6. 2 Related Work and Ideas WordNet’s taxonomic organization of nounsenses and verb-senses – in which very general categories are successively divided into increasingly informative sub-categories or instancelevel ideas – allows us to gauge the overlap in information content, and thus of meaning, of two lexical concepts. We need only identify the deepest point in the taxonomy at which this content starts to diverge. This point of divergence is often called the LCS, or least common subsumer, of two concepts (Pederson et al., 2004). Since sub-categories add new properties to those they inherit from their parents – Aristotle called these properties the differentia that stop a category system from trivially collapsing into itself – the depth of a lexical concept in a taxonomy is an intuitive proxy for its information content. Wu & Palmer (1994) use the depth of a lexical concept in the WordNet hierarchy as such a proxy, and thereby estimate the similarity of two lexical concepts as twice the depth of their LCS divided by the sum of their individual depths. Leacock and Chodorow (1998) instead use the length of the shortest path between two concepts as a proxy for the conceptual distance between them. To connect any two ideas in a hierarchical system, one must vertically ascend the hierarchy from one concept, change direction at a potential LCS, and then descend the hierarchy to reach the second concept. (Aristotle was also first to suggest this approach in his Poetics). Leacock and Chodorow normalize the length of this path by dividing its size (in nodes) by twice the depth of the deepest concept in the hierarchy; the latter is an upper bound on the distance between any two concepts in the hierarchy. Negating the log of this normalized length yields a corresponding similarity score. While the role of an LCS is merely implied in Leacock and Chodorow’s use of a shortest path, the LCS is pivotal nonetheless, and like that of Wu & Palmer, the approach uses an essentially vertical reasoning process to identify a single “best” generalization. Depth is a convenient proxy for information content, but more nuanced proxies can yield 661 more rounded similarity measures. Resnick (1995) draws on information theory to define the information content of a lexical concept as the negative log likelihood of its occurrence in a corpus, either explicitly (via a direct mention) or by presupposition (via a mention of any of its sub-categories or instances). Since the likelihood of a general category occurring in a corpus is higher than that of any of its sub-categories or instances, such categories are more predictable, and less informative, than rarer categories whose occurrences are less predictable and thus more informative. The negative log likelihood of the most informative LCS of two lexical concepts offers a reliable estimate of the amount of infor- mation shared by those concepts, and thus a good estimate of their similarity. Lin (1998) combines the intuitions behind Resnick’s metric and that of Wu and Palmer to estimate the similarity of two lexical concepts as an information ratio: twice the information content of their LCS divided by the sum of their individual information contents. Jiang and Conrath (1997) consider the converse notion of dissimilarity, noting that two lexical concepts are dissimilar to the extent that each contains information that is not shared by the other. So if the information content of their most informative LCS is a good measure of what they do share, then the sum of their individual information contents, minus twice the content of their most informative LCS, is a reliable estimate of their dissimilarity. Seco et al. (2006) presents a minor innovation, showing how Resnick’s notion of information content can be calculated without the use of an external corpus. Rather, when using Resnick’s metric (or that of Lin, or Jiang and Conrath) for measuring the similarity of lexical concepts in WordNet, one can use the category structure of WordNet itself to estimate infor- mation content. Typically, the more general a concept, the more descendants it will possess. Seco et al. thus estimate the information content of a lexical concept as the log of the sum of all its unique descendants (both direct and indirect), divided by the log of the total number of concepts in the entire hierarchy. Not only is this intrinsic view of information content convenient to use, without recourse to an external corpus, Seco et al. show that it offers a better estimate of information content than its extrinsic, corpus-based alternatives, as measured relative to average human similarity ratings for the 30 word-pairs in the Miller & Charles (1991) test set. A similarity measure can draw on other sources of information besides WordNet’s category structures. One might eke out additional information from WordNet’s textual glosses, as in Lesk (1986), or use category structures other than those offered by WordNet. Looking beyond WordNet, entries in the online encyclopedia Wikipedia are not only connected by a dense topology of lateral links, they are also organized by a rich hierarchy of overlapping categories. Strube and Ponzetto (2006) show how Wikipedia can support a measure of similarity (and relatedness) that better approximates human judgments than many WordNet-based measures. Nonetheless, WordNet can be a valuable component of a hybrid measure, and Agirre et al. (2009) use an SVM (support vector machine) to combine information from WordNet with information harvested from the web. Their best similarity measure achieves a remarkable 0.93 correlation with human judgments on the Miller & Charles word-pair set. Similarity is not always applied to pairs of concepts; it is sometimes analogically applied to pairs of pairs of concepts, as in proportional analogies of the form A is to B as C is to D (e.g., hacks are to writers as mercenaries are to soldiers, or chisels are to sculptors as scalpels are to surgeons). In such analogies, one is really assessing the similarity of the unstated relationship between each pair of concepts: thus, mercenaries are soldiers whose allegiance is paid for, much as hacks are writers with income-driven loyalties; sculptors use chisels to carve stone, while surgeons use scalpels to cut or carve flesh. Veale (2004) used WordNet to assess the similarity of A:B to C:D as a function of the combined similarity of A to C and of B to D. In contrast, Turney (2005) used the web to pursue a more divergent course, to represent the tacit relationships of A to B and of C to D as points in a highdimensional space. The dimensions of this space initially correspond to linking phrases on the web, before these dimensions are significantly reduced using singular value decomposition. In the infamous SAT test, an analogy A:B::C:D has four other pairs of concepts that serve as likely distractors (e.g. singer:songwriter for hack:writer) and the goal is to choose the most appropriate C:D pair for a given A:B pairing. Using variants of Wu and Palmer (1994) on the 374 SAT analogies of Turney (2005), Veale (2004) reports a success rate of 38–44% using only WordNet-based similarity. In contrast, Turney (2005) reports up to 55% success on the same analogies, partly because his approach aims 662 to match implicit relations rather than explicit concepts, and in part because it uses a divergent process to gather from the web as rich a perspec- tive as it can on these latent relationships. 2.1 Clever Comparisons Create Similarity Each of these approaches to similarity is a user of information, rather than a creator, and each fails to capture how a creative comparison (such as a metaphor) can spur a listener to view a topic from an atypical perspective. Camac & Glucksberg (1984) provide experimental evidence for the claim that “metaphors do not use preexisting associations to achieve their effects [… ] people use metaphors to create new relations between concepts.” They also offer a salutary reminder of an often overlooked fact: every comparison exploits information, but each is also a source of new information in its own right. Thus, “this cola is acid” reveals a different perspective on cola (e.g. as a corrosive substance or an irritating food) than “this acid is cola” highlights for acid (such as e.g., a familiar substance) Veale & Keane (1994) model the role of similarity in realizing the long-term perlocutionary effect of an informative comparison. For example, to compare surgeons to butchers is to encourage one to see all surgeons as more bloody, … crude or careless. The reverse comparison, of butchers to surgeons, encourages one to see butchers as more skilled and precise. Veale & Keane present a network model of memory, called Sapper, in which activation can spread between related concepts, thus allowing one concept to prime the properties of a neighbor. To interpret an analogy, Sapper lays down new activation-carrying bridges in memory between analogical counterparts, such as between surgeon & butcher, flesh & meat, and scalpel & cleaver. Comparisons can thus have lasting effects on how Sapper sees the world, changing the pattern of activation that arises when it primes a concept. Veale (2003) adopts a similarly dynamic view of similarity in WordNet, showing how an analogical comparison can result in the automatic addition of new categories and relations to WordNet itself. Veale considers the problem of finding an analogical mapping between different parts of WordNet’s noun-sense hierarchy, such as between instances of Greek god and Norse god, or between the letters of different alphabets, such as of Greek and Hebrew. But no structural similarity measure for WordNet exhibits enough discernment to e.g. assign a higher similarity to Zeus & Odin (each is the supreme deity of its pantheon) than to a pairing of Zeus and any other Norse god, just as no structural measure will assign a higher similarity to Alpha & Aleph or to Beta & Beth than to any random letter pairing. A fine-grained category hierarchy permits fine-grained similarity judgments, and though WordNet is useful, its sense hierarchies are not especially fine-grained. However, we can automatically make WordNet subtler and more discerning, by adding new fine-grained categories to unite lexical concepts whose similarity is not reflected by any existing categories. Veale (2003) shows how a property that is found in the glosses of two lexical concepts, of the same depth, can be combined with their LCS to yield a new fine-grained parent category, so e.g. “supreme” + deity = Supreme-deity (for Odin, Zeus, Jupiter, etc.) and “1 st” + letter = 1st-letter (for Alpha, Aleph, etc.) Selected aspects of the textual similarity of two WordNet glosses – the key to similarity in Lesk (1986) – can thus be reified into an explicitly categorical WordNet form. 3 Divergent (Re)Categorization To tap into a richer source of concept properties than WordNet’s glosses, we can use web ngrams. Consider these descriptions of a cowboy from the Google n-grams (Brants & Franz, 2006). The numbers to the right are Google frequency counts. a lonesome cowboy 432 a mounted cowboy 122 a grizzled cowboy 74 a swaggering cowboy 68 To find the stable properties that can underpin a meaningful fine-grained category for cowboy, we must seek out the properties that are so often presupposed to be salient of all cowboys that one can use them to anchor a simile, such as
5 0.44950294 344 acl-2013-The Effects of Lexical Resource Quality on Preference Violation Detection
Author: Jesse Dunietz ; Lori Levin ; Jaime Carbonell
Abstract: Lexical resources such as WordNet and VerbNet are widely used in a multitude of NLP tasks, as are annotated corpora such as treebanks. Often, the resources are used as-is, without question or examination. This practice risks missing significant performance gains and even entire techniques. This paper addresses the importance of resource quality through the lens of a challenging NLP task: detecting selectional preference violations. We present DAVID, a simple, lexical resource-based preference violation detector. With asis lexical resources, DAVID achieves an F1-measure of just 28.27%. When the resource entries and parser outputs for a small sample are corrected, however, the F1-measure on that sample jumps from 40% to 61.54%, and performance on other examples rises, suggesting that the algorithm becomes practical given refined resources. More broadly, this paper shows that resource quality matters tremendously, sometimes even more than algorithmic improvements.
6 0.4200798 302 acl-2013-Robust Automated Natural Language Processing with Multiword Expressions and Collocations
7 0.32936206 61 acl-2013-Automatic Interpretation of the English Possessive
8 0.3237215 3 acl-2013-A Comparison of Techniques to Automatically Identify Complex Words.
9 0.31705332 366 acl-2013-Understanding Verbs based on Overlapping Verbs Senses
10 0.31668171 95 acl-2013-Crawling microblogging services to gather language-classified URLs. Workflow and case study
11 0.3146916 65 acl-2013-BRAINSUP: Brainstorming Support for Creative Sentence Generation
12 0.31242979 53 acl-2013-Annotation of regular polysemy and underspecification
13 0.30429691 198 acl-2013-IndoNet: A Multilingual Lexical Knowledge Network for Indian Languages
14 0.30406108 268 acl-2013-PATHS: A System for Accessing Cultural Heritage Collections
15 0.30276179 322 acl-2013-Simple, readable sub-sentences
16 0.30251956 185 acl-2013-Identifying Bad Semantic Neighbors for Improving Distributional Thesauri
17 0.29714316 278 acl-2013-Patient Experience in Online Support Forums: Modeling Interpersonal Interactions and Medication Use
18 0.29073572 262 acl-2013-Offspring from Reproduction Problems: What Replication Failure Teaches Us
19 0.28947207 190 acl-2013-Implicatures and Nested Beliefs in Approximate Decentralized-POMDPs
20 0.27238268 236 acl-2013-Mapping Source to Target Strings without Alignment by Analogical Learning: A Case Study with Transliteration
topicId topicWeight
[(0, 0.056), (5, 0.348), (6, 0.017), (11, 0.041), (15, 0.016), (24, 0.042), (26, 0.036), (28, 0.015), (35, 0.069), (42, 0.044), (48, 0.037), (56, 0.014), (70, 0.044), (88, 0.036), (90, 0.024), (95, 0.054)]
simIndex simValue paperId paperTitle
same-paper 1 0.78914583 88 acl-2013-Computational considerations of comparisons and similes
Author: Vlad Niculae ; Victoria Yaneva
Abstract: This paper presents work in progress towards automatic recognition and classification of comparisons and similes. Among possible applications, we discuss the place of this task in text simplification for readers with Autism Spectrum Disorders (ASD), who are known to have deficits in comprehending figurative language. We propose an approach to comparison recognition through the use of syntactic patterns. Keeping in mind the requirements of autistic readers, we discuss the properties relevant for distinguishing semantic criteria like figurativeness and abstractness.
2 0.68817788 366 acl-2013-Understanding Verbs based on Overlapping Verbs Senses
Author: Kavitha Rajan
Abstract: Natural language can be easily understood by everyone irrespective of their differences in age or region or qualification. The existence of a conceptual base that underlies all natural languages is an accepted claim as pointed out by Schank in his Conceptual Dependency (CD) theory. Inspired by the CD theory and theories in Indian grammatical tradition, we propose a new set of meaning primitives in this paper. We claim that this new set of primitives captures the meaning inherent in verbs and help in forming an inter-lingual and computable ontological classification of verbs. We have identified seven primitive overlapping verb senses which substantiate our claim. The percentage of coverage of these primitives is 100% for all verbs in Sanskrit and Hindi and 3750 verbs in English. 1
3 0.67322648 3 acl-2013-A Comparison of Techniques to Automatically Identify Complex Words.
Author: Matthew Shardlow
Abstract: Identifying complex words (CWs) is an important, yet often overlooked, task within lexical simplification (The process of automatically replacing CWs with simpler alternatives). If too many words are identified then substitutions may be made erroneously, leading to a loss of meaning. If too few words are identified then those which impede a user’s understanding may be missed, resulting in a complex final text. This paper addresses the task of evaluating different methods for CW identification. A corpus of sentences with annotated CWs is mined from Simple Wikipedia edit histories, which is then used as the basis for several experiments. Firstly, the corpus design is explained and the results of the validation experiments using human judges are reported. Experiments are carried out into the CW identification techniques of: simplifying everything, frequency thresholding and training a support vector machine. These are based upon previous approaches to the task and show that thresholding does not perform significantly differently to the more na¨ ıve technique of simplifying everything. The support vector machine achieves a slight increase in precision over the other two methods, but at the cost of a dramatic trade off in recall.
4 0.3825171 46 acl-2013-An Infinite Hierarchical Bayesian Model of Phrasal Translation
Author: Trevor Cohn ; Gholamreza Haffari
Abstract: Modern phrase-based machine translation systems make extensive use of wordbased translation models for inducing alignments from parallel corpora. This is problematic, as the systems are incapable of accurately modelling many translation phenomena that do not decompose into word-for-word translation. This paper presents a novel method for inducing phrase-based translation units directly from parallel data, which we frame as learning an inverse transduction grammar (ITG) using a recursive Bayesian prior. Overall this leads to a model which learns translations of entire sentences, while also learning their decomposition into smaller units (phrase-pairs) recursively, terminating at word translations. Our experiments on Arabic, Urdu and Farsi to English demonstrate improvements over competitive baseline systems.
5 0.38240364 83 acl-2013-Collective Annotation of Linguistic Resources: Basic Principles and a Formal Model
Author: Ulle Endriss ; Raquel Fernandez
Abstract: Crowdsourcing, which offers new ways of cheaply and quickly gathering large amounts of information contributed by volunteers online, has revolutionised the collection of labelled data. Yet, to create annotated linguistic resources from this data, we face the challenge of having to combine the judgements of a potentially large group of annotators. In this paper we investigate how to aggregate individual annotations into a single collective annotation, taking inspiration from the field of social choice theory. We formulate a general formal model for collective annotation and propose several aggregation methods that go beyond the commonly used majority rule. We test some of our methods on data from a crowdsourcing experiment on textual entailment annotation.
6 0.38217622 85 acl-2013-Combining Intra- and Multi-sentential Rhetorical Parsing for Document-level Discourse Analysis
7 0.38181531 267 acl-2013-PARMA: A Predicate Argument Aligner
8 0.38177764 254 acl-2013-Multimodal DBN for Predicting High-Quality Answers in cQA portals
9 0.38165164 2 acl-2013-A Bayesian Model for Joint Unsupervised Induction of Sentiment, Aspect and Discourse Representations
10 0.3814477 318 acl-2013-Sentiment Relevance
11 0.38139823 17 acl-2013-A Random Walk Approach to Selectional Preferences Based on Preference Ranking and Propagation
12 0.38122594 172 acl-2013-Graph-based Local Coherence Modeling
13 0.38091731 272 acl-2013-Paraphrase-Driven Learning for Open Question Answering
14 0.38089585 187 acl-2013-Identifying Opinion Subgroups in Arabic Online Discussions
15 0.38064373 134 acl-2013-Embedding Semantic Similarity in Tree Kernels for Domain Adaptation of Relation Extraction
16 0.38060847 174 acl-2013-Graph Propagation for Paraphrasing Out-of-Vocabulary Words in Statistical Machine Translation
17 0.38027239 252 acl-2013-Multigraph Clustering for Unsupervised Coreference Resolution
18 0.38001266 224 acl-2013-Learning to Extract International Relations from Political Context
19 0.37968644 178 acl-2013-HEADY: News headline abstraction through event pattern clustering
20 0.37910128 250 acl-2013-Models of Translation Competitions