acl acl2012 acl2012-82 knowledge-graph by maker-knowledge-mining
Source: pdf
Author: Meni Adler ; Jonathan Berant ; Ido Dagan
Abstract: We present a novel text exploration model, which extends the scope of state-of-the-art technologies by moving from standard concept-based exploration to statement-based exploration. The proposed scheme utilizes the textual entailment relation between statements as the basis of the exploration process. A user of our system can explore the result space of a query by drilling down/up from one statement to another, according to entailment relations specified by an entailment graph and an optional concept taxonomy. As a prominent use case, we apply our exploration system and illustrate its benefit on the health-care domain. To the best of our knowledge this is the first implementation of an exploration system at the statement level that is based on the textual entailment relation. 1
Reference: text
sentIndex sentText sentNum sentScore
1 i lj l l Abstract We present a novel text exploration model, which extends the scope of state-of-the-art technologies by moving from standard concept-based exploration to statement-based exploration. [sent-10, score-0.912]
2 The proposed scheme utilizes the textual entailment relation between statements as the basis of the exploration process. [sent-11, score-0.931]
3 A user of our system can explore the result space of a query by drilling down/up from one statement to another, according to entailment relations specified by an entailment graph and an optional concept taxonomy. [sent-12, score-1.318]
4 As a prominent use case, we apply our exploration system and illustrate its benefit on the health-care domain. [sent-13, score-0.446]
5 To the best of our knowledge this is the first implementation of an exploration system at the statement level that is based on the textual entailment relation. [sent-14, score-0.876]
6 Standard search engines output a set of documents for a given query, but do not allow any exploration of the thematic structure in the retrieved information. [sent-16, score-0.547]
7 Thus, the need for tools that allow to effectively sift through a target set of documents is becoming ever more important. [sent-17, score-0.065]
8 Faceted search (Stoica and Hearst, 2007; K a¨ki, 2005) supports a better understanding of a target domain, by allowing exploration of data according to multiple views or facets. [sent-18, score-0.469]
9 For example, given a set of documents on Nobel Prize laureates we might have different facets corresponding to the laureate’s nationality, the year when the prize was awarded, the 79 field in which it was awarded, etc. [sent-19, score-0.136]
10 However, this type of exploration is still severely limited insofar that it only allows exploration by topic rather than content. [sent-20, score-0.856]
11 Put differently, we can only explore according to what a document is about rather than what a document actually says. [sent-21, score-0.076]
12 For instance, the facets for the query ‘asthma’ in the faceted search engine Yippy include the concepts allergy and children, but do not specify what are the exact relations between these concepts and the query (e. [sent-22, score-0.547]
13 (2010) proposed an exploration scheme that focuses on relations between concepts, which are derived from a graph describing textual entailment relations between propositions. [sent-26, score-1.089]
14 In their setting a proposition consists of a predicate with two arguments that are possibly replaced by variables, such as ‘X control asthma ’. [sent-27, score-0.859]
15 A graph that specifies an entailment relation ‘X control asthma → X affect antsathilmmae ’n can help a user, nwtrhool aiss browsing Xdo acf-uments dealing with substances that affect asthma, drill down and explore only substances that control asthma. [sent-28, score-1.55]
16 This type of exploration can be viewed as an extension of faceted search, where the new facet concentrates on the actual statements expressed in the texts. [sent-29, score-0.619]
17 ’s proposal, and present a novel entailment-based text exploration system, which we applied to the health-care domain. [sent-31, score-0.428]
18 A user of this system can explore the result space of her query, by drilling down/up from one proposition to another, according to a set of entailment relations described by an entailment graph. [sent-32, score-1.114]
19 In Figure 1, for example, the user looks for ‘things’ Proce Jedijung, sR oefpu thbeli c50 othf K Aonrneua,a8l -M14e Jtiunlgy o 2f0 t1h2e. [sent-33, score-0.08]
20 c s 2o0c1ia2ti Aosns fo cria Ctio nm fpourta Ctoiomnpault Laitniognuaislt Licisn,g puaigsteiscs 79–84, Figure 1: Exploring asthma results. [sent-35, score-0.594]
21 She invokes an ‘asthma’ query and starts drilling down the entailment graph to ‘X control asthma ’ (left column). [sent-37, score-1.316]
22 In order to examine the arguments of a selected proposition, the user may drill down/up a concept taxonomy that classifies terms that occur as arguments. [sent-38, score-0.501]
23 The user in Figure 1, for instance, drills down the concept taxonomy (middle column), in order to focus on Hormones that control asthma, such as ‘prednisone ’ (right column). [sent-39, score-0.381]
24 Each drill down/up induces a subset of the documents that correspond to the aforementioned selections. [sent-40, score-0.209]
25 The retrieved document in Figure 1 (bottom) is highlighted by the relevant proposition, which clearly states that prednisone is often given to treat asthma (and indeed in the entailment graph ‘X treat asthma ’ entails ‘X control asthma ’). [sent-41, score-2.599]
26 Our system is built over a corpus of documents, a set of propositions extracted from the documents, an entailment graph describing entailment relations between propositions, and, optionally, a concept hierarchy. [sent-42, score-1.212]
27 The system implementation for the healthcare domain, for instance, is based on a web-crawled health-care corpus, the propositions automatically 80 extracted from the corpus, entailment graphs borrowed from Berant et al. [sent-43, score-0.621]
28 To the best of our knowledge this is the first implementation of an exploration system, at the proposition level, based on the textual entailment relation. [sent-45, score-0.96]
29 In exploratory search, users are presented with a result set and a set of exploratory facets, which are proposals for refinements of the query that can lead to more focused sets of documents. [sent-48, score-0.233]
30 The user proceeds in the exploration of the document set by selecting specific documents (to read them) or by selecting specific facets, to refine the result set. [sent-50, score-0.621]
31 gov/ re search/uml s / Early exploration technologies were based on a single hierarchical conceptual clustering of information (Hofmann, 1999), enabling the user to drill up and down the concept hierarchies. [sent-54, score-0.754]
32 Hierarchical faceted meta-data (Stoica and Hearst, 2007), or faceted search, proposed more sophisticated exploration possibilities by providing multiple facets and a hierarchy per facet or dimension of the domain. [sent-55, score-0.752]
33 These types of exploration techniques were found to be useful for effective access of information (K¨ aki, 2005). [sent-56, score-0.428]
34 In this work, we suggest proposition-based exploration as an extension to concept-based exploration. [sent-57, score-0.428]
35 Our intuition is that text exploration can profit greatly from representing information not only at the level of individual concepts, but also at the propositional level, where the relations that link concepts to one another are represented effectively in a hierarchical entailment graph. [sent-58, score-1.092]
36 For example, ‘Levalbuterol is used to control various kinds of asthma ’ entails ‘Levalbuterol affects asthma ’. [sent-62, score-1.299]
37 In this paper, we use the notion of proposition to denote a specific type of text fragments, composed of a predicate with two arguments (e. [sent-63, score-0.185]
38 Textual entailment systems are often based on entailment rules which specify a directional inference relation between two fragments. [sent-66, score-0.798]
39 In this work, we focus on leveraging a common type of entailment rules, in which the left-hand-side of the rule (LHS) and the right-hand-side of the rule (RHS) are propositional templates - a proposition, where one or both of the arguments are replaced by a variable, e. [sent-67, score-0.674]
40 n propositional templates of a given corpus can be represented by an entailment graph (Berant et al. [sent-71, score-0.711]
41 The nodes of an entailment graph correspond to propositional templates, and its edges correspond to entailment relations (rules) between them. [sent-73, score-1.16]
42 Entailment graph representation is somewhat analogous to the formation of ontological relations between concepts of a given domain, where in our case the nodes 81 correspond to propositional templates rather than to concepts. [sent-74, score-0.492]
43 3 Exploration Model In this section we extend the scope of state-of-theart exploration technologies by moving from standard concept-based exploration to proposition-based exploration, or equivalently, statement-based exploration. [sent-75, score-0.912]
44 In our model, it is the entailment relation between propositional templates which determines the granularity of the viewed information space. [sent-76, score-0.616]
45 We first describe the inputs to the system and then detail our proposed exploration scheme. [sent-77, score-0.451]
46 The propositions are usually produced by an extraction method, such as TextRunner (Banko et al. [sent-81, score-0.156]
47 In order to support the exploration process, the documents are indexed by the propositional templates and argument terms of the extracted propositions. [sent-84, score-0.737]
48 Entailment graph for predicates The nodes of the entailment graph are propositional templates, where edges indicate entailment relations between templates (Section 2. [sent-85, score-1.303]
49 In order to avoid circularity in the exploration process, the graph is transformed into a DAG, by merging ‘equivalent’ nodes that are in the same strong connectivity component (as suggested by Berant et al. [sent-87, score-0.572]
50 Figure 2 illustrates the result of applying this procedure to a fragment of the entailment graph for ‘asthma ’ (i. [sent-90, score-0.504]
51 , for propositional templates with ‘asthma’ as one of the arguments). [sent-92, score-0.207]
52 Taxonomy for arguments The optional concept taxonomy maps terms to one or more pre-defined concepts, arranged in a hierarchical structure. [sent-93, score-0.339]
53 These terms may appear in the corpus as arguments of predicates. [sent-94, score-0.078]
54 Figure 3, for instance, illustrates a simple medical taxonomy, composed of three concepts (medical, diseases, drugs) and four terms (cancer, asthma, aspirin, flexeril). [sent-95, score-0.157]
55 Figure 2: Fragment of the entailment graph for ‘asthma’ (top), and its conversion to a DAG (bottom). [sent-96, score-0.504]
56 2 Exploration Scheme The objective ofthe exploration scheme is to support querying and offer facets for result exploration, in a visual manner. [sent-98, score-0.54]
57 The following components cover the various aspects of this objective, given the above system inputs: Querying The user enters a search term as a query, e. [sent-99, score-0.121]
58 The given term induces a subgraph of the entailment graph that contains all propositional templates (graph nodes) with which this term appears as an argument in the extracted propositions (see Figure 2). [sent-102, score-0.971]
59 1, where all nodes that have no parent are defined as the roots of the DAG. [sent-104, score-0.074]
60 Exploration process The user selects one of the entailment graph nodes (e. [sent-107, score-0.613]
61 At each exploration step, the user can drill down to a more specific template or drill up to a 82 Figure 3: Partial medical taxonomy. [sent-110, score-0.88]
62 Figure 4: The roots of the entailment graph for the ‘asthma ’ query. [sent-112, score-0.549]
63 more general template, by moving along the entailment hierarchy. [sent-113, score-0.424]
64 For example, the user in Figure 5, expands the root ‘associate X with asthma ’, in order to drill down through ‘X affect asthma’ to ‘X control Asthma ’. [sent-114, score-0.905]
65 Selecting a propositional template (Figure 1, left column) displays a concept taxonomy for the arguments that correspond to the variable in the selected template (Figure 1, middle column). [sent-115, score-0.556]
66 The user can explore these argument concepts by drilling up and down the concept taxonomy. [sent-116, score-0.409]
67 For example, in Figure 1 the user, who selected ‘X control Asthma ’, explores the arguments of this template by drilling down the taxonomy to the concept ‘Hormone ’. [sent-117, score-0.516]
68 Selecting a concept opens a third column, which lists the terms mapped to this concept that occurred as arguments of the selected template. [sent-118, score-0.306]
69 For example, in Figure 1, the user is examining the list of arguments for the template ‘X control Asthma’, which are mapped to the concept ‘Hormone ’, focusing on the argument ‘prednisone ’. [sent-119, score-0.451]
70 Figure 5: Part of the entailment graph for the ‘asthma ’ query, after two exploration steps. [sent-120, score-0.932]
71 Document retrieval At any stage, the list of documents induced by the current selected template, concept and argument is presented to the user, where in each document snippet the relevant proposition components are highlighted. [sent-122, score-0.339]
72 The highlighted extraction in the snippet, ‘prednisone treat asthma ’, entails the proposition selected during exploration, ‘prednisone control asthma ’. [sent-124, score-1.442]
73 The search service implements full-text and faceted search, and document indexing. [sent-126, score-0.21]
74 The entailment service handles the logic of the entailment relations (for both the entailment graph and the taxonomy). [sent-130, score-1.378]
75 The index server applies periodic indexing of new texts, and the exploration server serves the exploration application on querying, exploration, and data 83 Figure 6: Block diagram of the exploration system. [sent-131, score-1.379]
76 The exploration application is the front-end user application for the whole exploration process described above (Section 3. [sent-133, score-0.996]
77 5 Application to the Health-care Domain As a prominent use case, we applied our exploration system to the health-care domain. [sent-135, score-0.446]
78 With the advent of the internet and social media, patients now have access to new sources of medical information: consumer health articles, forums, and social networks (Boulos and Wheeler, 2007). [sent-136, score-0.16]
79 A typical non-expert health information searcher is uncertain about her exact questions and is unfamiliar with medical terminology (Trivedi, 2009). [sent-137, score-0.16]
80 Exploring relevant information about a given medical issue can be essential and time-critical. [sent-138, score-0.078]
81 System implementation For the search service, we used SolR servlet, where the data service is built over FTP. [sent-139, score-0.076]
82 The exploration application is implemented as a web application. [sent-140, score-0.458]
83 The texts deal with various aspects of the health care domain: answers to questions, surveys on diseases, articles on life-style, etc. [sent-142, score-0.1]
84 We extracted propositions from the health-care corpus, by applying the method described by Berant et al. [sent-143, score-0.176]
85 The corpus was parsed, and propositions were extracted from dependency trees according to the method suggested by Lin and Pantel (2001), where propositions are dependency paths between two arguments of a predi- cate. [sent-145, score-0.41]
86 We filtered out any proposition where one of the arguments is not a term mapped to a medical concept in the UMLS taxonomy. [sent-146, score-0.388]
87 For the entailment graph we used the 23 entailment graphs published by Berant et al. [sent-147, score-0.914]
88 For the argument taxonomy we employed UMLS a database that maps natural language phrases to over one million unique concept identifiers (CUIs) in the healthcare domain. [sent-149, score-0.294]
89 The CUIs are also mapped in UMLS to a concept taxonomy for the health-care domain. [sent-150, score-0.243]
90 14 8 : 8 0 8 0 / expl orat ion – 6 Conclusion and Future Work We presented a novel exploration model, which extends the scope of state-of-the-art exploration technologies by moving from standard concept-based exploration to proposition-based exploration. [sent-154, score-1.34]
91 Our model combines the textual entailment paradigm within the exploration process, with application to the health-care domain. [sent-155, score-0.883]
92 According to our model, it is the entailment relation between propositions, en- coded by the entailment graph and the taxonomy, which leads the user between more specific and more general statements throughout the search result space. [sent-156, score-1.072]
93 We believe that employing the entailment relation between propositions, which focuses on the statements expressed in the documents, can contribute to the exploration field and improve information access. [sent-157, score-0.896]
94 Our current application to the health-care domain relies on a small set of entailment graphs for 23 medical concepts. [sent-158, score-0.539]
95 Our ongoing research focuses on the challenging task of learning a larger entailment graph for the health-care domain. [sent-159, score-0.525]
96 We are also investigating methods for evaluating the exploration process (Borlund and Ingwersen, 1997). [sent-160, score-0.428]
97 As noted by Qu and Furnas (2008), the success of an exploratory search system does not depend simply on how many relevant documents will be retrieved for a given query, but more broadly on how well the system helps the user with the exploratory process. [sent-161, score-0.381]
98 0 social software: an enabling suite of sociable technologies in health and health care education. [sent-185, score-0.182]
99 Findex: search result categories help users when document ranking fails. [sent-202, score-0.068]
100 Model-driven formative evaluation of exploratory search: A study under a sensemaking framework. [sent-212, score-0.091]
wordName wordTfidf (topN-words)
[('asthma', 0.594), ('exploration', 0.428), ('entailment', 0.389), ('propositions', 0.156), ('propositional', 0.135), ('drill', 0.122), ('taxonomy', 0.118), ('graph', 0.115), ('berant', 0.11), ('proposition', 0.107), ('faceted', 0.107), ('concept', 0.103), ('exploratory', 0.091), ('drilling', 0.087), ('prednisone', 0.087), ('health', 0.082), ('control', 0.08), ('user', 0.08), ('concepts', 0.079), ('arguments', 0.078), ('medical', 0.078), ('templates', 0.072), ('facets', 0.064), ('dag', 0.056), ('levalbuterol', 0.052), ('stoica', 0.052), ('query', 0.051), ('template', 0.05), ('dagan', 0.048), ('facet', 0.046), ('roots', 0.045), ('documents', 0.044), ('israel', 0.043), ('umls', 0.042), ('search', 0.041), ('relations', 0.04), ('argument', 0.038), ('statements', 0.038), ('column', 0.037), ('textual', 0.036), ('allergy', 0.035), ('borlund', 0.035), ('boulos', 0.035), ('cuis', 0.035), ('healthcare', 0.035), ('hormone', 0.035), ('ilan', 0.035), ('onatha', 0.035), ('ramat', 0.035), ('substances', 0.035), ('moving', 0.035), ('service', 0.035), ('retrieved', 0.034), ('entails', 0.031), ('awarded', 0.03), ('aviv', 0.03), ('application', 0.03), ('nodes', 0.029), ('affect', 0.029), ('ido', 0.029), ('querying', 0.028), ('prize', 0.028), ('diseases', 0.028), ('tel', 0.028), ('document', 0.027), ('gan', 0.026), ('bottom', 0.025), ('fader', 0.025), ('subgraph', 0.025), ('bar', 0.023), ('diagram', 0.023), ('statement', 0.023), ('inputs', 0.023), ('mapped', 0.022), ('banko', 0.022), ('correspond', 0.022), ('explore', 0.022), ('server', 0.021), ('selecting', 0.021), ('domain', 0.021), ('focuses', 0.021), ('graphs', 0.021), ('hierarchical', 0.021), ('scope', 0.021), ('becoming', 0.021), ('handles', 0.021), ('hearst', 0.021), ('induces', 0.021), ('extracted', 0.02), ('snippet', 0.02), ('qu', 0.02), ('relation', 0.02), ('scheme', 0.02), ('optional', 0.019), ('edges', 0.019), ('ijcai', 0.019), ('prominent', 0.018), ('care', 0.018), ('highlighted', 0.018), ('treat', 0.018)]
simIndex simValue paperId paperTitle
same-paper 1 1.0000002 82 acl-2012-Entailment-based Text Exploration with Application to the Health-care Domain
Author: Meni Adler ; Jonathan Berant ; Ido Dagan
Abstract: We present a novel text exploration model, which extends the scope of state-of-the-art technologies by moving from standard concept-based exploration to statement-based exploration. The proposed scheme utilizes the textual entailment relation between statements as the basis of the exploration process. A user of our system can explore the result space of a query by drilling down/up from one statement to another, according to entailment relations specified by an entailment graph and an optional concept taxonomy. As a prominent use case, we apply our exploration system and illustrate its benefit on the health-care domain. To the best of our knowledge this is the first implementation of an exploration system at the statement level that is based on the textual entailment relation. 1
2 0.22460704 80 acl-2012-Efficient Tree-based Approximation for Entailment Graph Learning
Author: Jonathan Berant ; Ido Dagan ; Meni Adler ; Jacob Goldberger
Abstract: Learning entailment rules is fundamental in many semantic-inference applications and has been an active field of research in recent years. In this paper we address the problem of learning transitive graphs that describe entailment rules between predicates (termed entailment graphs). We first identify that entailment graphs exhibit a “tree-like” property and are very similar to a novel type of graph termed forest-reducible graph. We utilize this property to develop an iterative efficient approximation algorithm for learning the graph edges, where each iteration takes linear time. We compare our approximation algorithm to a recently-proposed state-of-the-art exact algorithm and show that it is more efficient and scalable both theoretically and empirically, while its output quality is close to that given by the optimal solution of the exact algorithm.
3 0.16651335 72 acl-2012-Detecting Semantic Equivalence and Information Disparity in Cross-lingual Documents
Author: Yashar Mehdad ; Matteo Negri ; Marcello Federico
Abstract: We address a core aspect of the multilingual content synchronization task: the identification of novel, more informative or semantically equivalent pieces of information in two documents about the same topic. This can be seen as an application-oriented variant of textual entailment recognition where: i) T and H are in different languages, and ii) entailment relations between T and H have to be checked in both directions. Using a combination of lexical, syntactic, and semantic features to train a cross-lingual textual entailment system, we report promising results on different datasets.
4 0.1458938 65 acl-2012-Crowdsourcing Inference-Rule Evaluation
Author: Naomi Zeichner ; Jonathan Berant ; Ido Dagan
Abstract: The importance of inference rules to semantic applications has long been recognized and extensive work has been carried out to automatically acquire inference-rule resources. However, evaluating such resources has turned out to be a non-trivial task, slowing progress in the field. In this paper, we suggest a framework for evaluating inference-rule resources. Our framework simplifies a previously proposed “instance-based evaluation” method that involved substantial annotator training, making it suitable for crowdsourcing. We show that our method produces a large amount of annotations with high inter-annotator agreement for a low cost at a short period of time, without requiring training expert annotators.
5 0.13566287 36 acl-2012-BIUTEE: A Modular Open-Source System for Recognizing Textual Entailment
Author: Asher Stern ; Ido Dagan
Abstract: This paper introduces BIUTEE1 , an opensource system for recognizing textual entailment. Its main advantages are its ability to utilize various types of knowledge resources, and its extensibility by which new knowledge resources and inference components can be easily integrated. These abilities make BIUTEE an appealing RTE system for two research communities: (1) researchers of end applications, that can benefit from generic textual inference, and (2) RTE researchers, who can integrate their novel algorithms and knowledge resources into our system, saving the time and effort of developing a complete RTE system from scratch. Notable assistance for these re- searchers is provided by a visual tracing tool, by which researchers can refine and “debug” their knowledge resources and inference components.
6 0.12664238 53 acl-2012-Combining Textual Entailment and Argumentation Theory for Supporting Online Debates Interactions
7 0.087825567 78 acl-2012-Efficient Search for Transformation-based Inference
8 0.057626069 184 acl-2012-String Re-writing Kernel
9 0.054778744 64 acl-2012-Crosslingual Induction of Semantic Roles
10 0.04850968 44 acl-2012-CSNIPER - Annotation-by-query for Non-canonical Constructions in Large Corpora
11 0.046308823 208 acl-2012-Unsupervised Relation Discovery with Sense Disambiguation
12 0.045147162 147 acl-2012-Modeling the Translation of Predicate-Argument Structure for SMT
13 0.044103604 142 acl-2012-Mining Entity Types from Query Logs via User Intent Modeling
14 0.043354712 180 acl-2012-Social Event Radar: A Bilingual Context Mining and Sentiment Analysis Summarization System
15 0.043214958 159 acl-2012-Pattern Learning for Relation Extraction with a Hierarchical Topic Model
16 0.041504018 212 acl-2012-Using Search-Logs to Improve Query Tagging
17 0.040439453 133 acl-2012-Learning to "Read Between the Lines" using Bayesian Logic Programs
18 0.038989272 112 acl-2012-Humor as Circuits in Semantic Networks
19 0.038890805 206 acl-2012-UWN: A Large Multilingual Lexical Knowledge Base
20 0.038254477 191 acl-2012-Temporally Anchored Relation Extraction
topicId topicWeight
[(0, -0.111), (1, 0.058), (2, -0.025), (3, 0.074), (4, 0.007), (5, 0.068), (6, -0.036), (7, 0.221), (8, 0.022), (9, 0.037), (10, -0.061), (11, 0.266), (12, 0.082), (13, -0.188), (14, -0.007), (15, -0.065), (16, 0.007), (17, -0.005), (18, 0.078), (19, 0.006), (20, -0.09), (21, 0.063), (22, 0.029), (23, 0.006), (24, 0.064), (25, -0.039), (26, -0.035), (27, -0.011), (28, 0.021), (29, -0.039), (30, -0.018), (31, -0.015), (32, 0.191), (33, 0.013), (34, -0.042), (35, -0.176), (36, 0.022), (37, 0.038), (38, 0.017), (39, 0.102), (40, 0.081), (41, -0.047), (42, 0.13), (43, -0.016), (44, 0.005), (45, -0.055), (46, 0.047), (47, 0.091), (48, 0.118), (49, -0.083)]
simIndex simValue paperId paperTitle
same-paper 1 0.97366887 82 acl-2012-Entailment-based Text Exploration with Application to the Health-care Domain
Author: Meni Adler ; Jonathan Berant ; Ido Dagan
Abstract: We present a novel text exploration model, which extends the scope of state-of-the-art technologies by moving from standard concept-based exploration to statement-based exploration. The proposed scheme utilizes the textual entailment relation between statements as the basis of the exploration process. A user of our system can explore the result space of a query by drilling down/up from one statement to another, according to entailment relations specified by an entailment graph and an optional concept taxonomy. As a prominent use case, we apply our exploration system and illustrate its benefit on the health-care domain. To the best of our knowledge this is the first implementation of an exploration system at the statement level that is based on the textual entailment relation. 1
2 0.79248863 80 acl-2012-Efficient Tree-based Approximation for Entailment Graph Learning
Author: Jonathan Berant ; Ido Dagan ; Meni Adler ; Jacob Goldberger
Abstract: Learning entailment rules is fundamental in many semantic-inference applications and has been an active field of research in recent years. In this paper we address the problem of learning transitive graphs that describe entailment rules between predicates (termed entailment graphs). We first identify that entailment graphs exhibit a “tree-like” property and are very similar to a novel type of graph termed forest-reducible graph. We utilize this property to develop an iterative efficient approximation algorithm for learning the graph edges, where each iteration takes linear time. We compare our approximation algorithm to a recently-proposed state-of-the-art exact algorithm and show that it is more efficient and scalable both theoretically and empirically, while its output quality is close to that given by the optimal solution of the exact algorithm.
3 0.76981306 72 acl-2012-Detecting Semantic Equivalence and Information Disparity in Cross-lingual Documents
Author: Yashar Mehdad ; Matteo Negri ; Marcello Federico
Abstract: We address a core aspect of the multilingual content synchronization task: the identification of novel, more informative or semantically equivalent pieces of information in two documents about the same topic. This can be seen as an application-oriented variant of textual entailment recognition where: i) T and H are in different languages, and ii) entailment relations between T and H have to be checked in both directions. Using a combination of lexical, syntactic, and semantic features to train a cross-lingual textual entailment system, we report promising results on different datasets.
4 0.67375231 53 acl-2012-Combining Textual Entailment and Argumentation Theory for Supporting Online Debates Interactions
Author: Elena Cabrio ; Serena Villata
Abstract: Blogs and forums are widely adopted by online communities to debate about various issues. However, a user that wants to cut in on a debate may experience some difficulties in extracting the current accepted positions, and can be discouraged from interacting through these applications. In our paper, we combine textual entailment with argumentation theory to automatically extract the arguments from debates and to evaluate their acceptability.
5 0.6522122 65 acl-2012-Crowdsourcing Inference-Rule Evaluation
Author: Naomi Zeichner ; Jonathan Berant ; Ido Dagan
Abstract: The importance of inference rules to semantic applications has long been recognized and extensive work has been carried out to automatically acquire inference-rule resources. However, evaluating such resources has turned out to be a non-trivial task, slowing progress in the field. In this paper, we suggest a framework for evaluating inference-rule resources. Our framework simplifies a previously proposed “instance-based evaluation” method that involved substantial annotator training, making it suitable for crowdsourcing. We show that our method produces a large amount of annotations with high inter-annotator agreement for a low cost at a short period of time, without requiring training expert annotators.
6 0.39052019 36 acl-2012-BIUTEE: A Modular Open-Source System for Recognizing Textual Entailment
7 0.3271924 133 acl-2012-Learning to "Read Between the Lines" using Bayesian Logic Programs
8 0.30535734 77 acl-2012-Ecological Evaluation of Persuasive Messages Using Google AdWords
9 0.28863478 164 acl-2012-Private Access to Phrase Tables for Statistical Machine Translation
10 0.26092511 42 acl-2012-Bootstrapping via Graph Propagation
11 0.25896037 78 acl-2012-Efficient Search for Transformation-based Inference
12 0.2457152 160 acl-2012-Personalized Normalization for a Multilingual Chat System
13 0.23990497 184 acl-2012-String Re-writing Kernel
14 0.21157344 156 acl-2012-Online Plagiarized Detection Through Exploiting Lexical, Syntax, and Semantic Information
15 0.21098793 44 acl-2012-CSNIPER - Annotation-by-query for Non-canonical Constructions in Large Corpora
16 0.1997941 212 acl-2012-Using Search-Logs to Improve Query Tagging
17 0.1997252 186 acl-2012-Structuring E-Commerce Inventory
18 0.19848837 219 acl-2012-langid.py: An Off-the-shelf Language Identification Tool
19 0.19385396 142 acl-2012-Mining Entity Types from Query Logs via User Intent Modeling
20 0.18623638 215 acl-2012-WizIE: A Best Practices Guided Development Environment for Information Extraction
topicId topicWeight
[(25, 0.019), (26, 0.035), (28, 0.017), (30, 0.041), (37, 0.019), (39, 0.044), (82, 0.018), (84, 0.029), (85, 0.46), (90, 0.074), (92, 0.071), (94, 0.012), (99, 0.05)]
simIndex simValue paperId paperTitle
same-paper 1 0.89893162 82 acl-2012-Entailment-based Text Exploration with Application to the Health-care Domain
Author: Meni Adler ; Jonathan Berant ; Ido Dagan
Abstract: We present a novel text exploration model, which extends the scope of state-of-the-art technologies by moving from standard concept-based exploration to statement-based exploration. The proposed scheme utilizes the textual entailment relation between statements as the basis of the exploration process. A user of our system can explore the result space of a query by drilling down/up from one statement to another, according to entailment relations specified by an entailment graph and an optional concept taxonomy. As a prominent use case, we apply our exploration system and illustrate its benefit on the health-care domain. To the best of our knowledge this is the first implementation of an exploration system at the statement level that is based on the textual entailment relation. 1
2 0.84287357 13 acl-2012-A Graphical Interface for MT Evaluation and Error Analysis
Author: Meritxell Gonzalez ; Jesus Gimenez ; Lluis Marquez
Abstract: Error analysis in machine translation is a necessary step in order to investigate the strengths and weaknesses of the MT systems under development and allow fair comparisons among them. This work presents an application that shows how a set of heterogeneous automatic metrics can be used to evaluate a test bed of automatic translations. To do so, we have set up an online graphical interface for the ASIYA toolkit, a rich repository of evaluation measures working at different linguistic levels. The current implementation of the interface shows constituency and dependency trees as well as shallow syntactic and semantic annotations, and word alignments. The intelligent visualization of the linguistic structures used by the metrics, as well as a set of navigational functionalities, may lead towards advanced methods for automatic error analysis.
3 0.81769371 152 acl-2012-Multilingual WSD with Just a Few Lines of Code: the BabelNet API
Author: Roberto Navigli ; Simone Paolo Ponzetto
Abstract: In this paper we present an API for programmatic access to BabelNet a wide-coverage multilingual lexical knowledge base and multilingual knowledge-rich Word Sense Disambiguation (WSD). Our aim is to provide the research community with easy-to-use tools to perform multilingual lexical semantic analysis and foster further research in this direction. – –
4 0.80438495 165 acl-2012-Probabilistic Integration of Partial Lexical Information for Noise Robust Haptic Voice Recognition
Author: Khe Chai Sim
Abstract: This paper presents a probabilistic framework that combines multiple knowledge sources for Haptic Voice Recognition (HVR), a multimodal input method designed to provide efficient text entry on modern mobile devices. HVR extends the conventional voice input by allowing users to provide complementary partial lexical information via touch input to improve the efficiency and accuracy of voice recognition. This paper investigates the use of the initial letter of the words in the utterance as the partial lexical information. In addition to the acoustic and language models used in automatic speech recognition systems, HVR uses the haptic and partial lexical models as additional knowledge sources to reduce the recognition search space and suppress confusions. Experimental results show that both the word error rate and runtime factor can be re- duced by a factor of two using HVR.
5 0.77735752 96 acl-2012-Fast and Robust Part-of-Speech Tagging Using Dynamic Model Selection
Author: Jinho D. Choi ; Martha Palmer
Abstract: This paper presents a novel way of improving POS tagging on heterogeneous data. First, two separate models are trained (generalized and domain-specific) from the same data set by controlling lexical items with different document frequencies. During decoding, one of the models is selected dynamically given the cosine similarity between each sentence and the training data. This dynamic model selection approach, coupled with a one-pass, leftto-right POS tagging algorithm, is evaluated on corpora from seven different genres. Even with this simple tagging algorithm, our system shows comparable results against other state-of-the-art systems, and gives higher accuracies when evaluated on a mixture of the data. Furthermore, our system is able to tag about 32K tokens per second. this model selection approach to more sophisticated tagging improve their robustness even We believe that can be applied algorithms and further.
6 0.67223197 162 acl-2012-Post-ordering by Parsing for Japanese-English Statistical Machine Translation
7 0.50789869 206 acl-2012-UWN: A Large Multilingual Lexical Knowledge Base
8 0.48959559 158 acl-2012-PORT: a Precision-Order-Recall MT Evaluation Metric for Tuning
9 0.48694378 29 acl-2012-Assessing the Effect of Inconsistent Assessors on Summarization Evaluation
10 0.47184241 52 acl-2012-Combining Coherence Models and Machine Translation Evaluation Metrics for Summarization Evaluation
11 0.46949291 136 acl-2012-Learning to Translate with Multiple Objectives
12 0.45711103 24 acl-2012-A Web-based Evaluation Framework for Spatial Instruction-Giving Systems
13 0.45480269 132 acl-2012-Learning the Latent Semantics of a Concept from its Definition
14 0.45475468 72 acl-2012-Detecting Semantic Equivalence and Information Disparity in Cross-lingual Documents
15 0.45473123 214 acl-2012-Verb Classification using Distributional Similarity in Syntactic and Semantic Structures
16 0.44667229 138 acl-2012-LetsMT!: Cloud-Based Platform for Do-It-Yourself Machine Translation
17 0.44223464 48 acl-2012-Classifying French Verbs Using French and English Lexical Resources
18 0.43668041 219 acl-2012-langid.py: An Off-the-shelf Language Identification Tool
19 0.43376318 63 acl-2012-Cross-lingual Parse Disambiguation based on Semantic Correspondence
20 0.43209335 46 acl-2012-Character-Level Machine Translation Evaluation for Languages with Ambiguous Word Boundaries