acl acl2010 acl2010-35 knowledge-graph by maker-knowledge-mining
Source: pdf
Author: Konstantina Garoufi ; Alexander Koller
Abstract: We present a natural language generation approach which models, exploits, and manipulates the non-linguistic context in situated communication, using techniques from AI planning. We show how to generate instructions which deliberately guide the hearer to a location that is convenient for the generation of simple referring expressions, and how to generate referring expressions with context-dependent adjectives. We implement and evaluate our approach in the framework of the Challenge on Generating Instructions in Virtual Environments, finding that it performs well even under the constraints of realtime generation.
Reference: text
sentIndex sentText sentNum sentScore
1 Automated planning for situated natural language generation Konstantina Garoufi and Alexander Koller Cluster of Excellence “Multimodal Computing and Interaction” Saarland University, Saarbr¨ ucken, Germany {garoufi , kol ler} @mmci . [sent-1, score-0.458]
2 de Abstract We present a natural language generation approach which models, exploits, and manipulates the non-linguistic context in situated communication, using techniques from AI planning. [sent-3, score-0.239]
3 We show how to generate instructions which deliberately guide the hearer to a location that is convenient for the generation of simple referring expressions, and how to generate referring expressions with context-dependent adjectives. [sent-4, score-0.507]
4 1 Introduction The problem of situated natural language generation (NLG)—i. [sent-6, score-0.196]
5 Consider the following segment of discourse between an instruction giver (IG) and an instruction follower (IF), which is adapted from the SCARE corpus (Stoia et al. [sent-13, score-0.296]
6 In this example, the IG plans to refer to an object (here, a button); and in order to do so, gives a navigation instruction to guide the IF to a convenient location at which she can then use a simple referring expression (RE). [sent-17, score-0.305]
7 That is, there is an interaction between navigation instructions (intended to manipulate the non-linguistic context in a certain way) and referring expressions (which exploit the non-linguistic context). [sent-18, score-0.379]
8 This paper presents an approach to generation which is able to model the effect of an utterance on the non-linguistic context, and to intentionally generate utterances such as the above as part of a process of referring to objects. [sent-20, score-0.238]
9 Our approach builds upon the CRISP generation system (Koller and Stone, 2007), which translates generation problems into planning problems and solves these with an AI planner. [sent-21, score-0.446]
10 We extend the CRISP planning operators with the perlocutionary effects that uttering a particular word has on the physical environment if it is understood correctly; more specifically, on the position and orientation of the hearer. [sent-22, score-0.654]
11 A second contribution of our paper is the generation of REs involving context-dependent adjectives: A button can be described as “the left blue 1573 ProceedingUsp opfs thaela 4, 8Stwhe Adnennu,a 1l1- M16ee Jtiunlgy o 2f0 t1h0e. [sent-25, score-0.491]
12 c ss2o0c1ia0ti Aosnso focria Ctioonm fpourta Ctoiomnpault Laitniognuaislt Licisn,g puaigsetisc 1s573–1582, button” even if there is a red button to its left. [sent-27, score-0.454]
13 Thus, unlike most other RE generation approaches, we are not restricted to building an RE by simply intersecting lexically specified sets representing the extensions of different attributes, but can correctly generate expressions whose meaning depends on the context in a number of ways. [sent-29, score-0.203]
14 Next to the practical usefulness this evaluation establishes, we argue that our approach to jointly modeling the grammatical and physical effects of a communicative action can also inform new models of the pragmatics of speech acts. [sent-34, score-0.312]
15 First, it exploits the situated communicative setting to deliberately modify the context in which an RE is generated. [sent-42, score-0.259]
16 Such a component, however, can only inform a system on whether to choose navigation over RE generation at a given point of the discourse, and is not able to help it decide what kind of navigational instructions to generate so that subsequent REs become simple. [sent-51, score-0.387]
17 Unlike van Deemter, we integrate the RE generation process tightly with the syntactic realization, which allows us to generate REs with more than one context-dependent modifier and model the effect of their linear order on the meaning of the phrase. [sent-53, score-0.161]
18 The research we report here can be seen as combining Appelt’s idea of using planning for sentence-level NLG with a computationally benign variant of Perrault et al. [sent-60, score-0.262]
19 ’s approach of modeling the intended perlocutionary effects of a speech act as the effects of a planning operator. [sent-61, score-0.483]
20 Our work is linked to a growing body of very recent work that applies modern planning research to various problems in NLG (Steedman and Petrick, 2007; Brenner and Kruijff-Korbayov a´, 2008; Benotti, 2009). [sent-62, score-0.262]
21 , 2010b), that is, as an instruction giving system for virtual worlds. [sent-68, score-0.195]
22 3 Sentence generation as planning Our work is based on the CRISP system (Koller and Stone, 2007), which encodes sentence generation with tree-adjoining grammars (TAG; (Joshi and Schabes, 1997)) as an AI planning problem and solves that using efficient planners. [sent-70, score-0.708]
23 1 TAG sentence generation The CRISP generation problem (like that of SPUD (Stone et al. [sent-75, score-0.184]
24 Let’s say, for example, that we have a knowledge base {push (e, j,b1) , John (j) , button (b1) , button (b2) , r{epdu (b1) }. [sent-84, score-0.624]
25 1b; we can read off the output sentence “John pushes the red button” from the leaves of the derived tree we build in this way. [sent-88, score-0.21]
26 2 TAG generation as planning In the CRISP system, Koller and Stone (2007) show how this generation problem can be solved by converting it into a planning problem (Nau et al. [sent-90, score-0.708]
27 The basic idea is to encode the partial derivation in the planning state, and to encode the action of adding each elementary tree in the planning operators. [sent-92, score-0.7]
28 The encoding of our example as a planning problem is shown in Fig. [sent-93, score-0.262]
29 In the example, we start with an initial state which contains the entire knowledge base, plus atoms subst(S, root) and ref(root, e) expressing that we want to generate a sentence about the event e. [sent-95, score-0.214]
30 ¬John (y) → ¬distractor(u, y) the-button(u, x): Precond: subst(NP, u) , ref(u, x) , button (x) Effect: ¬subst(NP, u) , canadjoin (N, u) , ∀¬ysu. [sent-99, score-0.391]
31 ¬bsbtu(tNtoPn, ,(uy)) →ca n¬addijsotirna(cNto,ru(u),, y) red(u, x) : Precond: canadjoin (N, u) , ref(u, x) , red (x) Effect: ∀y. [sent-100, score-0.221]
32 ¬red (y) → ¬distractor(u, y) Figure 2: CRISP planning operators for the elementary trees in Fig. [sent-101, score-0.435]
33 The planning state maintains information about which individual each node refers to in the ref atoms. [sent-104, score-0.4]
34 1 Finally, the action introduces a number of distractor atoms including distractor(n2, e) and distractor(n2, b2), expressing that the RE at n2 can still be misunderstood by the hearer as e or b2. [sent-106, score-0.492]
35 In this new state, all subst and distractor atoms for n1 can be eliminated with the action John(n1 ,j). [sent-107, score-0.82]
36 Now because the action the-button also introduced the atom canadjoin(N, n2), we can remove the fi- nal distractor atom by applying red(n2, b1). [sent-110, score-0.467]
37 Goal states in CRISP planning problems are characterized by axioms such as ∀A∀u. [sent-112, score-0.262]
38 3 Decoding the plan An AI planner such as FF (Hoffmann and Nebel, 2001) can compute a plan for a planning problem that consists of the planning operators in Fig. [sent-119, score-0.789]
39 The basic idea of this decoding step is that an action with a precondition subst(A, u) fills the substitution node u, while an action with a precondition canadjoin (A, u) adjoins into a node of category A in the elementary tree that was substituted into u. [sent-123, score-0.492]
40 The instruction follower (IF), who is located on the map at position pos3,2 facing north, sees the scene from the first-person perspective as in Fig. [sent-129, score-0.195]
41 Now an instruction giver (IG) could instruct the IF to press the button b1 in this scene by saying “push the button on the wall to your left”. [sent-131, score-0.851]
42 1 Situated CRISP We define a lexicon for SCRISP to be a CRISP lexicon in which every lexicon entry may also describe non-linguistic conditions, non-linguistic effects and imperative effects. [sent-142, score-0.272]
43 Figure 5: SCRISP planning operators for the lexicon in Fig. [sent-167, score-0.443]
44 In addition, the lexicon entry for “turn left” specifies that, under the assumption that the IF understands and follows the instruction, they will turn 90 degrees to the left after hearing it. [sent-171, score-0.164]
45 The planning operators are written in a way that assumes that the intended (perlocutionary) effects of an utterance actually come true. [sent-172, score-0.47]
46 This assumption is crucial in connecting the non-linguistic effects of one SCRISP action to the non-linguistic preconditions of another, and generalizes to a scalable model of planning perlocutionary acts. [sent-173, score-0.553]
47 We then translate a SCRISP generation problem into a planning problem. [sent-176, score-0.354]
48 In addition to what CRISP does, we translate all non-linguistic conditions into preconditions and all non-linguistic effects into effects of the planning operator, adding any free variables to the operator’s parameters. [sent-177, score-0.434]
49 Finally, we add information about the situated environment to the initial state, and specify the planning goal by adding to–do(P) atoms for each atom P that is to be placed on the IF’s agenda. [sent-182, score-0.55]
50 2 An example Now let’s look at how this generates the appropri- ate instructions for our example scene of Fig. [sent-184, score-0.189]
51 We encode the state of the world as depicted in the map in an initial state which contains, among others, the atoms player–pos(pos3,2), player–ori(north), next–ori–left(north, west), 1577 visible(pos3,2, west, b1), etc. [sent-186, score-0.25]
52 Next to the ordinary grammatical effects from CRISP, this action makes player–ori(west) true. [sent-189, score-0.178]
53 The new state does not contain any subst atoms, but we can continue the sentence by adjoining “and”, i. [sent-190, score-0.373]
54 Because turnleft changed the player orientation, the visible precondition of push is now satisfied too (unlike in the initial state, in which b1 was not visible). [sent-194, score-0.468]
55 Applying the action push now introduces the need to substitute a noun phrase for the object, which we can eliminate with an application of the-button(n2, b1) as in Subsection 3. [sent-195, score-0.23]
56 Since there are no other visible buttons from pos3,2 facing west, there are no remaining distractor atoms at this point, and a goal state has been reached. [sent-197, score-0.525]
57 Together, this four-step plan decodes into the sentence “turn left and push the button”. [sent-198, score-0.29]
58 The final state contains the atoms to–do(push (b1)) and to–do(turnleft), indicating that an IF that understands and accepts this instruction also accepts these two commitments into their to-do list. [sent-199, score-0.324]
59 3 instead of b1, say with the instruction “push the left button”. [sent-201, score-0.235]
60 Figure 6: SCRISP operators for contextdependent and context-independent adjectives. [sent-221, score-0.164]
61 depend on the meaning of the phrase they modify: “the left button” is not necessarily both a button and further to the left than all other objects, it is only the leftmost object among the buttons. [sent-222, score-0.486]
62 Our encoding of context-dependent adjectives as planning operators is shown in Fig. [sent-226, score-0.478]
63 We only show the operators here for lack of space; they can of course be computed automatically from lexicon entries. [sent-228, score-0.181]
64 In addition to the ordinary CRISP precon- ditions, the left operator has a precondition requiring that no current distractor for the RE u is to the left of x, capturing a presupposition of the adjective. [sent-229, score-0.572]
65 Its effect is that everything that is to the right of x is no longer a distractor for u. [sent-230, score-0.284]
66 Notice that we allow that there may still be distractors after left has been applied (above or below x); we only require unique reference in the goal state. [sent-231, score-0.161]
67 The state after these two actions is not a goal state, because it still contains the atom distractor(n1 , b3) (the plant f1 was removed as a distractor by the action the-button). [sent-235, score-0.459]
68 have modeled the spatial objects in the initial state atoms; in particular, we Then the action instance left(n1 , b2) is applicable in this state, as there is no other object that is still a distractor in this state and that is to the left of b2. [sent-237, score-0.621]
69 Thus we have reached a goal state; the complete plan decodes to the sentence “push the left button”. [sent-239, score-0.168]
70 This system is sensitive to the order in which operators for context-dependent adjectives are applied. [sent-240, score-0.216]
71 To generate the RE “the upper left button”, for instance, we first apply the left action and then the upper action, and therefore upper only needs to remove distractors in the leftmost position. [sent-241, score-0.484]
72 These action sequences succeed in removing all distractors for different context states, which is consistent with the difference in meaning between the two REs. [sent-243, score-0.225]
73 Furthermore, notice that the adjective operators themselves do not interact directly with the encoding of the context in atoms like visible and player–pos, just like the noun operators in Section 4 didn’t. [sent-244, score-0.582]
74 The REs to which the adjectives and nouns contribute are introduced by verb operators; it is these verb operators that inspect the current context and initialize the distractor set for the new RE appropriately. [sent-245, score-0.512]
75 This makes the correctness of the generated sentence independent of the order in which noun and adjective operators occur in the plan. [sent-246, score-0.174]
76 the red left button” is rather odd even when it is a semantically correct description, whereas “the left red button” is fine. [sent-259, score-0.458]
77 In our lexicon we assign adjectives denoting spatial relations (“left”) to one class and adjectives denoting color (“red”) to another; then we require that spatial adjectives must always precede color adjectives. [sent-261, score-0.407]
78 We enforce this by keeping track ofthe currentpremodifier index ofthe RE in atoms of the form premod–index. [sent-262, score-0.173]
79 6 illustrate, color adjectives such as “red” have index one and can only be used while the index is not higher; once an adjective from a higher class (such as “left”, of a class with index two) is used, the premod–index precondition of the “red” operator will fail. [sent-265, score-0.381]
80 For this reason, we can generate a plan for “the left red button”, but not for “? [sent-266, score-0.309]
81 In this challenge, systems must generate real-time instructions that help users perform a task in a treasure-hunt virtual environment such as the one shown in Fig. [sent-295, score-0.227]
82 As in the challenge, the task is considered as solved if the player has correctly been led through manipulating all target objects required to discover and collect the treasure; in World 2, the minimum number of such targets is eight. [sent-304, score-0.162]
83 An RE is successfully resolved if it results in the manipulation of the referent, whereas manipulation of an alarm-triggering distractor ends the game unsuccessfully. [sent-305, score-0.313]
84 Then, for each of the communicative goals, it generates instructions using SCRISP, segments them into navigation and action parts, and presents these to the user as separate instructions sequentially (see Table 1). [sent-313, score-0.541]
85 We use the FF planner (Hoffmann and Nebel, 2001 ; Koller and Hoffmann, 2010) to solve the planning problems. [sent-315, score-0.305]
86 The maximum planning time for any instruction is 1. [sent-316, score-0.41]
87 So although our planning-based system tackles a very difficult search problem, FF is very good at solving it—fast enough to generate instructions in real time. [sent-319, score-0.18]
88 We hand-coded a correct distinguishing RE for each target button in the world; the only way in which Baseline A reacts to changes of the context is to describe on which wall the button is with respect to the user’s current orientation (e. [sent-322, score-0.702]
89 “Press the right red button on the wall to your right”). [sent-324, score-0.454]
90 Baseline B, like the original “Austin” system, issues navigation instructions by precomputing the shortest path from the IF’s current location to the target, and generates REs using the description logic based algorithm of Areces et al. [sent-331, score-0.222]
91 Unlike the original system, which inflexibly navigates the user all the way to the target, Baseline B starts off with navigation, and opportunistically instructs the IF to push a button once it has become visible and can be described by a distinguishing RE. [sent-333, score-0.53]
92 Specifically, a button cannot be referred to as “the right red button” if it is not the rightmost of all visible objects—which explains the long chain of navigational instructions the system produced in Table 1. [sent-336, score-0.727]
93 7 Conclusion In this paper, we have shown how situated instructions can be generated using AI planning. [sent-344, score-0.246]
94 We exploited the planner’s ability to model the perlocutionary effects of communicative actions for efficient generation. [sent-345, score-0.22]
95 We showed how this made it possible to generate instructions that manipulate the non-linguistic context in convenient ways, and to generate correct REs with context-dependent adjectives. [sent-346, score-0.298]
96 In addition, we plan to experiment with assigning costs to planning operators in a metric planning problem (Hoffmann, 2002) in order to model the cognitive cost of an RE (Krahmer et al. [sent-349, score-0.704]
97 On a more theoretical level, the SCRISP actions model the physical effects of a correctly understood and grounded instruction directly as effects of the planning operator. [sent-351, score-0.586]
98 This is computationally much less complex than classical speech act planning (Perrault and Allen, 1980), in which the intended physical effect comes at the end of a long chain of inferences. [sent-352, score-0.329]
99 Computational interpretations of the Gricean maxims in the generation of referring expressions. [sent-386, score-0.169]
100 The FF planning system: Fast plan generation through heuristic search. [sent-394, score-0.396]
wordName wordTfidf (topN-words)
[('subst', 0.328), ('button', 0.312), ('scrisp', 0.302), ('crisp', 0.263), ('planning', 0.262), ('distractor', 0.253), ('instruction', 0.148), ('instructions', 0.142), ('red', 0.142), ('operators', 0.138), ('atoms', 0.131), ('res', 0.127), ('push', 0.122), ('precond', 0.118), ('koller', 0.118), ('player', 0.116), ('action', 0.108), ('situated', 0.104), ('re', 0.102), ('nlg', 0.096), ('visible', 0.096), ('ref', 0.093), ('generation', 0.092), ('stone', 0.089), ('left', 0.087), ('perlocutionary', 0.081), ('precondition', 0.081), ('navigation', 0.08), ('canadjoin', 0.079), ('adjectives', 0.078), ('referring', 0.077), ('ori', 0.075), ('distractors', 0.074), ('effects', 0.07), ('communicative', 0.069), ('pushes', 0.068), ('ig', 0.066), ('stoia', 0.066), ('krahmer', 0.058), ('hoffmann', 0.056), ('west', 0.054), ('atom', 0.053), ('premod', 0.053), ('spud', 0.053), ('turnleft', 0.053), ('self', 0.051), ('virtual', 0.047), ('scene', 0.047), ('perrault', 0.046), ('objects', 0.046), ('state', 0.045), ('lexicon', 0.043), ('context', 0.043), ('deliberately', 0.043), ('planner', 0.043), ('plan', 0.042), ('index', 0.042), ('root', 0.04), ('ff', 0.04), ('imperative', 0.039), ('brenner', 0.039), ('decodes', 0.039), ('scare', 0.039), ('generate', 0.038), ('spatial', 0.037), ('manipulate', 0.037), ('adjective', 0.036), ('physical', 0.036), ('orientation', 0.035), ('elementary', 0.035), ('navigational', 0.035), ('entry', 0.034), ('deemter', 0.034), ('derivation', 0.033), ('generating', 0.033), ('operator', 0.032), ('referent', 0.032), ('rg', 0.032), ('instruct', 0.032), ('preconditions', 0.032), ('presupposition', 0.032), ('uttering', 0.032), ('effect', 0.031), ('np', 0.031), ('challenge', 0.031), ('upper', 0.03), ('intersecting', 0.03), ('byron', 0.03), ('manipulation', 0.03), ('world', 0.029), ('pragmatics', 0.029), ('color', 0.028), ('donna', 0.028), ('kees', 0.027), ('alexander', 0.027), ('areces', 0.026), ('contextdependent', 0.026), ('darla', 0.026), ('dornhege', 0.026), ('firms', 0.026)]
simIndex simValue paperId paperTitle
same-paper 1 0.99999994 35 acl-2010-Automated Planning for Situated Natural Language Generation
Author: Konstantina Garoufi ; Alexander Koller
Abstract: We present a natural language generation approach which models, exploits, and manipulates the non-linguistic context in situated communication, using techniques from AI planning. We show how to generate instructions which deliberately guide the hearer to a location that is convenient for the generation of simple referring expressions, and how to generate referring expressions with context-dependent adjectives. We implement and evaluate our approach in the framework of the Challenge on Generating Instructions in Virtual Environments, finding that it performs well even under the constraints of realtime generation.
2 0.16413406 168 acl-2010-Learning to Follow Navigational Directions
Author: Adam Vogel ; Dan Jurafsky
Abstract: We present a system that learns to follow navigational natural language directions. Where traditional models learn from linguistic annotation or word distributions, our approach is grounded in the world, learning by apprenticeship from routes through a map paired with English descriptions. Lacking an explicit alignment between the text and the reference path makes it difficult to determine what portions of the language describe which aspects of the route. We learn this correspondence with a reinforcement learning algorithm, using the deviation of the route we follow from the intended path as a reward signal. We demonstrate that our system successfully grounds the meaning of spatial terms like above and south into geometric properties of paths.
3 0.15380006 202 acl-2010-Reading between the Lines: Learning to Map High-Level Instructions to Commands
Author: S.R.K. Branavan ; Luke Zettlemoyer ; Regina Barzilay
Abstract: In this paper, we address the task of mapping high-level instructions to sequences of commands in an external environment. Processing these instructions is challenging—they posit goals to be achieved without specifying the steps required to complete them. We describe a method that fills in missing information using an automatically derived environment model that encodes states, transitions, and commands that cause these transitions to happen. We present an efficient approximate approach for learning this environment model as part of a policygradient reinforcement learning algorithm for text interpretation. This design enables learning for mapping high-level instructions, which previous statistical methods cannot handle.1
4 0.097848795 92 acl-2010-Don't 'Have a Clue'? Unsupervised Co-Learning of Downward-Entailing Operators.
Author: Cristian Danescu-Niculescu-Mizil ; Lillian Lee
Abstract: Researchers in textual entailment have begun to consider inferences involving downward-entailing operators, an interesting and important class of lexical items that change the way inferences are made. Recent work proposed a method for learning English downward-entailing operators that requires access to a high-quality collection of negative polarity items (NPIs). However, English is one of the very few languages for which such a list exists. We propose the first approach that can be applied to the many languages for which there is no pre-existing high-precision database of NPIs. As a case study, we apply our method to Romanian and show that our method yields good results. Also, we perform a cross-linguistic analysis that suggests interesting connections to some findings in linguistic typology.
5 0.091336571 231 acl-2010-The Prevalence of Descriptive Referring Expressions in News and Narrative
Author: Raquel Hervas ; Mark Finlayson
Abstract: Generating referring expressions is a key step in Natural Language Generation. Researchers have focused almost exclusively on generating distinctive referring expressions, that is, referring expressions that uniquely identify their intended referent. While undoubtedly one of their most important functions, referring expressions can be more than distinctive. In particular, descriptive referring expressions those that provide additional information not required for distinction are critical to flu– – ent, efficient, well-written text. We present a corpus analysis in which approximately one-fifth of 7,207 referring expressions in 24,422 words ofnews and narrative are descriptive. These data show that if we are ever to fully master natural language generation, especially for the genres of news and narrative, researchers will need to devote more attention to understanding how to generate descriptive, and not just distinctive, referring expressions. 1 A Distinctive Focus Generating referring expressions is a key step in Natural Language Generation (NLG). From early treatments in seminal papers by Appelt (1985) and Reiter and Dale (1992) to the recent set of Referring Expression Generation (REG) Challenges (Gatt et al., 2009) through different corpora available for the community (Eugenio et al., 1998; van Deemter et al., 2006; Viethen and Dale, 2008), generating referring expressions has become one of the most studied areas of NLG. Researchers studying this area have, almost without exception, focused exclusively on how to generate distinctive referring expressions, that is, referring expressions that unambiguously idenMark Alan Finlayson Computer Science and Artificial Intelligence Laboratory Massachusetts Institute of Technology Cambridge, MA, 02139 USA markaf@mit .edu tify their intended referent. Referring expressions, however, may be more than distinctive. It is widely acknowledged that they can be used to achieve multiple goals, above and beyond distinction. Here we focus on descriptive referring expressions, that is, referring expressions that are not only distinctive, but provide additional information not required for identifying their intended referent. Consider the following text, in which some of the referring expressions have been underlined: Once upon a time there was a man, who had three daughters. They lived in a house and their dresses were made of fabric. While a bit strange, the text is perfectly wellformed. All the referring expressions are distinctive, in that we can properly identify the referents of each expression. But the real text, the opening lines to the folktale The Beauty and the Beast, is actually much more lyrical: Once upon a time there was a rich merchant, who had three daughters. They lived in a very fine house and their gowns were made of the richest fabric sewn with jewels. All the boldfaced portions namely, the choice of head nouns, the addition of adjectives, the use of appositive phrases serve to perform a descriptive function, and, importantly, are all unnecessary for distinction! In all of these cases, the author is using the referring expressions as a vehicle for communicating information about the referents. This descriptive information is sometimes – – new, sometimes necessary for understanding the text, and sometimes just for added flavor. But when the expression is descriptive, as opposed to distinctive, this additional information is not required for identifying the referent of the expression, and it is these sorts of referring expressions that we will be concerned with here. 49 Uppsala,P Srwoce de dni,n 1g1s- 1of6 t Jhuely AC 20L1 20 .1 ?0c 2 C0o1n0fe Aresnsoceci Sathio rnt f Poarp Ceorsm,p paugteastio 4n9a–l5 L4i,nguistics Although these sorts of referring expression have been mostly ignored by researchers in this area1 , we show in this corpus study that descriptive expressions are in fact quite prevalent: nearly one-fifth of referring expressions in news and narrative are descriptive. In particular, our data, the trained judgments of native English speakers, show that 18% of all distinctive referring expressions in news and 17% of those in narrative folktales are descriptive. With this as motivation, we argue that descriptive referring expressions must be studied more carefully, especially as the field progresses from referring in a physical, immediate context (like that in the REG Challenges) to generating more literary forms of text. 2 Corpus Annotation This is a corpus study; our procedure was therefore to define our annotation guidelines (Section 2.1), select texts to annotate (2.2), create an annotation tool for our annotators (2.3), and, finally, train annotators, have them annotate referring expressions’ constituents and function, and then adjudicate the double-annotated texts into a gold standard (2.4). 2.1 Definitions We wrote an annotation guide explaining the difference between distinctive and descriptive referring expressions. We used the guide when training annotators, and it was available to them while annotating. With limited space here we can only give an outline of what is contained in the guide; for full details see (Finlayson and Herv a´s, 2010a). Referring Expressions We defined referring expressions as referential noun phrases and their coreferential expressions, e.g., “John kissed Mary. She blushed.”. This included referring expressions to generics (e.g., “Lions are fierce”), dates, times, and numbers, as well as events if they were referred to using a noun phrase. We included in each referring expression all the determiners, quantifiers, adjectives, appositives, and prepositional phrases that syntactically attached to that expression. When referring expressions were nested, all the nested referring expressions were also marked separately. Nuclei vs. Modifiers In the only previous corpus study of descriptive referring expressions, on 1With the exception of a small amount of work, discussed in Section 4. museum labels, Cheng et al. (2001) noted that descriptive information is often integrated into referring expressions using modifiers to the head noun. To study this, and to allow our results to be more closely compared with Cheng’s, we had our annotators split referring expressions into their constituents, portions called either nuclei or modifiers. The nuclei were the portions of the referring expression that performed the ‘core’ referring function; the modifiers were those portions that could be varied, syntactically speaking, independently of the nuclei. Annotators then assigned a distinctive or descriptive function to each constituent, rather than the referring expression as a whole. Normally, the nuclei corresponded to the head of the noun phrase. In (1), the nucleus is the token king, which we have here surrounded with square brackets. The modifiers, surrounded by parentheses, are The and old. (1) (The) (old) [king] was wise. Phrasal modifiers were marked as single modifiers, for example, in (2). (2) (The) [roof] (of the house) collapsed. It is significant that we had our annotators mark and tag the nuclei of referring expressions. Cheng and colleagues only mentioned the possibility that additional information could be introduced in the modifiers. However, O’Donnell et al. (1998) observed that often the choice of head noun can also influence the function of a referring expression. Consider (3), in which the word villain is used to refer to the King. (3) The King assumed the throne today. I ’t trust (that) [villain] one bit. don The speaker could have merely used him to refer to the King–the choice of that particular head noun villain gives us additional information about the disposition of the speaker. Thus villain is descriptive. Function: Distinctive vs. Descriptive As already noted, instead of tagging the whole referring expression, annotators tagged each constituent (nuclei and modifiers) as distinctive or descriptive. The two main tests for determining descriptiveness were (a) if presence of the constituent was unnecessary for identifying the referent, or (b) if 50 the constituent was expressed using unusual or ostentatious word choice. If either was true, the constituent was considered descriptive; otherwise, it was tagged as distinctive. In cases where the constituent was completely irrelevant to identifying the referent, it was tagged as descriptive. For example, in the folktale The Princess and the Pea, from which (1) was extracted, there is only one king in the entire story. Thus, in that story, the king is sufficient for identification, and therefore the modifier old is descriptive. This points out the importance of context in determining distinctiveness or descriptiveness; if there had been a roomful of kings, the tags on those modifiers would have been reversed. There is some question as to whether copular predicates, such as the plumber in (4), are actually referring expressions. (4) John is the plumber Our annotators marked and tagged these constructions as normal referring expressions, but they added an additional flag to identify them as copular predicates. We then excluded these constructions from our final analysis. Note that copular predicates were treated differently from appositives: in appositives the predicate was included in the referring expression, and in most cases (again, depending on context) was marked descriptive (e.g., John, the plumber, slept.). 2.2 Text Selection Our corpus comprised 62 texts, all originally written in English, from two different genres, news and folktales. We began with 30 folktales of different sizes, totaling 12,050 words. These texts were used in a previous work on the influence of dialogues on anaphora resolution algorithms (Aggarwal et al., 2009); they were assembled with an eye toward including different styles, different authors, and different time periods. Following this, we matched, approximately, the number of words in the folktales by selecting 32 texts from Wall Street Journal section of the Penn Treebank (Marcus et al., 1993). These texts were selected at ran- dom from the first 200 texts in the corpus. 2.3 The Story Workbench We used the Story Workbench application (Finlayson, 2008) to actually perform the annotation. The Story Workbench is a semantic annotation program that, among other things, includes the ability to annotate referring expressions and coreferential relationships. We added the ability to annotate nuclei, modifiers, and their functions by writing a workbench “plugin” in Java that could be installed in the application. The Story Workbench is not yet available to the public at large, being in a limited distribution beta testing phase. The developers plan to release it as free software within the next year. At that time, we also plan to release our plugin as free, downloadable software. 2.4 Annotation & Adjudication The main task of the study was the annotation of the constituents of each referring expression, as well as the function (distinctive or descriptive) of each constituent. The system generated a first pass of constituent analysis, but did not mark functions. We hired two native English annotators, neither of whom had any linguistics background, who corrected these automatically-generated constituent analyses, and tagged each constituent as descriptive or distinctive. Every text was annotated by both annotators. Adjudication of the differences was conducted by discussion between the two annotators; the second author moderated these discussions and settled irreconcilable disagreements. We followed a “train-as-you-go” paradigm, where there was no distinct training period, but rather adjudication proceeded in step with annotation, and annotators received feedback during those sessions. We calculated two measures of inter-annotator agreement: a kappa statistic and an f-measure, shown in Table 1. All of our f-measures indicated that annotators agreed almost perfectly on the location of referring expressions and their breakdown into constituents. These agreement calculations were performed on the annotators’ original corrected texts. All the kappa statistics were calculated for two tags (nuclei vs. modifier for the constituents, and distinctive vs. descriptive for the functions) over both each token assigned to a nucleus or modifier and each referring expression pair. Our kappas indicate moderate to good agreement, especially for the folktales. These results are expected because of the inherent subjectivity of language. During the adjudication sessions it became clear that different people do not consider the same information 51 as obvious or descriptive for the same concepts, and even the contexts deduced by each annotators from the texts were sometimes substantially different. 3 Results Table 2 lists the primary results of the study. We considered a referring expression descriptive if any of its constituents were descriptive. Thus, 18% of the referring expressions in the corpus added additional information beyond what was required to unambiguously identify their referent. The results were similar in both genres. Tales Articles Total Texts303262 Words Sentences 12,050 904 12,372 571 24,422 1,475 Ref. Exp.3,6813,5267,207 Dist. Ref. Exp. 3,057 2,830 5,887 Desc. Ref. Exp. 609 672 1,281 % Dist. Ref.83%81%82% % Desc. Ref. 17% 19% Table 2: Primary results. 18% Table 3 contains the percentages of descriptive and distinctive tags broken down by constituent. Like Cheng’s results, our analysis shows that descriptive referring expressions make up a significant fraction of all referring expressions. Although Cheng did not examine nuclei, our results show that the use of descriptive nuclei is small but not negligible. 4 Relation to the Field Researchers working on generating referring expressions typically acknowledge that referring expressions can perform functions other than distinction. Despite this widespread acknowledgment, researchers have, for the most part, explicitly ignored these functions. Exceptions to this trend Tales Articles Total Nuclei3,6663,5027,168 Max. Nuc/Ref Dist. Nuc. 1 95% 1 97% 1 96% Desc. Nuc. 5% 3% 4% Modifiers2,2773,6275,904 Avg. Mod/Ref Max. Mod/Ref Dist. Mod. Desc. Mod. 0.6 4 78% 22% 1.0 6 81% 19% 0.8 6 80% 20% Table 3: Breakdown of Constituent Tags are three. First is the general study of aggregation in the process of referring expression generation. Second and third are corpus studies by Cheng et al. (2001) and Jordan (2000a) that bear on the prevalence of descriptive referring expressions. The NLG subtask of aggregation can be used to imbue referring expressions with a descriptive function (Reiter and Dale, 2000, §5.3). There is a specific nk (iRned otefr aggregation 0c0al0le,d § embedding t ihsa at moves information from one clause to another inside the structure of a separate noun phrase. This type of aggregation can be used to transform two sentences such as “The princess lived in a castle. She was pretty ” into “The pretty princess lived in a castle ”. The adjective pretty, previously a cop- ular predicate, becomes a descriptive modifier of the reference to the princess, making the second text more natural and fluent. This kind of aggregation is widely used by humans for making the discourse more compact and efficient. In order to create NLG systems with this ability, we must take into account the caveat, noted by Cheng (1998), that any non-distinctive information in a referring expression must not lead to confusion about the distinctive function of the referring expression. This is by no means a trivial problem this sort of aggregation interferes with referring and coherence planning at both a local and global level (Cheng and Mellish, 2000; Cheng et al., 2001). It is clear, from the current state of the art of NLG, that we have not yet obtained a deep enough understanding of aggregation to enable us to handle these interactions. More research on the topic is needed. Two previous corpus studies have looked at the use of descriptive referring expressions. The first showed explicitly that people craft descriptive referring expressions to accomplish different – 52 goals. Jordan and colleagues (Jordan, 2000b; Jordan, 2000a) examined the use of referring expressions using the COCONUT corpus (Eugenio et al., 1998). They tested how domain and discourse goals can influence the content of non-pronominal referring expressions in a dialogue context, checking whether or not a subject’s goals led them to include non-referring information in a referring expression. Their results are intriguing because they point toward heretofore unexamined constraints, utilities and expectations (possibly genre- or styledependent) that may underlie the use ofdescriptive information to perform different functions, and are not yet captured by aggregation modules in particular or NLG systems in general. In the other corpus study, which partially inspired this work, Cheng and colleagues analyzed a set of museum descriptions, the GNOME corpus (Poesio, 2004), for the pragmatic functions of referring expressions. They had three functions in their study, in contrast to our two. Their first function (marked by their uniq tag) was equiv- alent to our distinctive function. The other two were specializations of our descriptive tag, where they differentiated between additional information that helped to understand the text (int), or additional information not necessary for understanding (att r). Despite their annotators seeming to have trouble distinguishing between the latter two tags, they did achieve good overall inter-annotator agreement. They identified 1,863 modifiers to referring expressions in their corpus, of which 47.3% fulfilled a descriptive (att r or int) function. This is supportive of our main assertion, namely, that descriptive referring expressions, not only crucial for efficient and fluent text, are actually a significant phenomenon. It is interesting, though, that Cheng’s fraction of descriptive referring expression was so much higher than ours (47.3% versus our 18%). We attribute this substantial difference to genre, in that Cheng studied museum labels, in which the writer is spaceconstrained, having to pack a lot of information into a small label. The issue bears further study, and perhaps will lead to insights into differences in writing style that may be attributed to author or genre. 5 Contributions We make two contributions in this paper. First, we assembled, double-annotated, and adjudicated into a gold-standard a corpus of 24,422 words. We marked all referring expressions, coreferential relations, and referring expression constituents, and tagged each constituent as having a descriptive or distinctive function. We wrote an annotation guide and created software that allows the annotation of this information in free text. The corpus and the guide are available on-line in a permanent digital archive (Finlayson and Herv a´s, 2010a; Finlayson and Herv a´s, 2010b). The software will also be released in the same archive when the Story Workbench annotation application is released to the public. This corpus will be useful for the automatic generation and analysis of both descriptive and distinctive referring expressions. Any kind of system intended to generate text as humans do must take into account that identifica- tion is not the only function of referring expressions. Many analysis applications would benefit from the automatic recognition of descriptive referring expressions. Second, we demonstrated that descriptive referring expressions comprise a substantial fraction (18%) of the referring expressions in news and narrative. Along with museum descriptions, studied by Cheng, it seems that news and narrative are genres where authors naturally use a large number ofdescriptive referring expressions. Given that so little work has been done on descriptive referring expressions, this indicates that the field would be well served by focusing more attention on this phenomenon. Acknowledgments This work was supported in part by the Air Force Office of Scientific Research under grant number A9550-05-1-0321, as well as by the Office of Naval Research under award number N00014091059. Any opinions, findings, and con- clusions or recommendations expressed in this paper are those of the authors and do not necessarily reflect the views of the Office of Naval Research. This research is also partially funded the Spanish Ministry of Education and Science (TIN200914659-C03-01) and Universidad Complutense de Madrid (GR58/08). We also thank Whitman Richards, Ozlem Uzuner, Peter Szolovits, Patrick Winston, Pablo Gerv a´s, and Mark Seifter for their helpful comments and discussion, and thank our annotators Saam Batmanghelidj and Geneva Trotter. 53 References Alaukik Aggarwal, Pablo Gerv a´s, and Raquel Herv a´s. 2009. Measuring the influence of errors induced by the presence of dialogues in reference clustering of narrative text. In Proceedings of ICON-2009: 7th International Conference on Natural Language Processing, India. Macmillan Publishers. Douglas E. Appelt. 1985. Planning English referring expressions. Artificial Intelligence, 26: 1–33. Hua Cheng and Chris Mellish. 2000. Capturing the interaction between aggregation and text planning in two generation systems. In INLG ’00: First international conference on Natural Language Generation 2000, pages 186–193, Morristown, NJ, USA. Association for Computational Linguistics. Hua Cheng, Massimo Poesio, Renate Henschel, and Chris Mellish. 2001 . Corpus-based np modifier generation. In NAACL ’01: Second meeting of the North American Chapter of the Association for Computational Linguistics on Language technologies 2001, pages 1–8, Morristown, NJ, USA. Association for Computational Linguistics. Hua Cheng. 1998. Embedding new information into referring expressions. In ACL-36: Proceedings of the 36thAnnual Meeting ofthe Associationfor Computational Linguistics and 17th International Conference on Computational Linguistics, pages 1478– 1480, Morristown, NJ, USA. Association for Computational Linguistics. Barbara Di Eugenio, Johanna D. Moore, Pamela W. Jordan, and Richmond H. Thomason. 1998. An empirical investigation of proposals in collaborative dialogues. In Proceedings of the 17th international conference on Computational linguistics, pages 325–329, Morristown, NJ, USA. Association for Computational Linguistics. Mark A. Finlayson and Raquel Herv a´s. 2010a. Annotation guide for the UCM/MIT indications, referring expressions, and coreference corpus (UMIREC corpus). Technical Report MIT-CSAIL-TR-2010-025, MIT Computer Science and Artificial Intelligence Laboratory. http://hdl.handle.net/1721. 1/54765. Mark A. Finlayson and Raquel Herv a´s. 2010b. UCM/MIT indications, referring expressions, and coreference corpus (UMIREC corpus). Work product, MIT Computer Science and Artificial Intelligence Laboratory. http://hdl.handle.net/1721 .1/54766. Mark A. Finlayson. 2008. Collecting semantics in the wild: The Story Workbench. In Proceedings of the AAAI Fall Symposium on Naturally-Inspired Artificial Intelligence, pages 46–53, Menlo Park, CA, USA. AAAI Press. Albert Gatt, Anja Belz, and Eric Kow. 2009. The TUNA-REG challenge 2009: overview and evaluation results. In ENLG ’09: Proceedings of the 12th European Workshop on Natural Language Generation, pages 174–182, Morristown, NJ, USA. Association for Computational Linguistics. Pamela W. Jordan. 2000a. Can nominal expressions achieve multiple goals?: an empirical study. In ACL ’00: Proceedings of the 38th Annual Meeting on Association for Computational Linguistics, pages 142– 149, Morristown, NJ, USA. Association for Computational Linguistics. Pamela W. Jordan. 2000b. Influences on attribute selection in redescriptions: A corpus study. In Proceedings of CogSci2000, pages 250–255. Mitchell P. Marcus, Mary Ann Marcinkiewicz, and Beatrice Santorini. 1993. Building a large annotated corpus of english: the penn treebank. Computational Linguistics, 19(2):3 13–330. Michael O’Donnell, Hua Cheng, and Janet Hitzeman. 1998. Integrating referring and informing in NP planning. In Proceedings of COLING-ACL’98 Workshop on the Computational Treatment of Nominals, pages 46–56. Massimo Poesio. 2004. Discourse annotation and semantic annotation in the GNOME corpus. In DiscAnnotation ’04: Proceedings of the 2004 ACL Workshop on Discourse Annotation, pages 72–79, Morristown, NJ, USA. Association for Computational Linguistics. Ehud Reiter and Robert Dale. 1992. A fast algorithm for the generation of referring expressions. In Proceedings of the 14th conference on Computational linguistics, Nantes, France. Ehud Reiter and Robert Dale. 2000. Building Natural Language Generation Systems. Cambridge University Press. Kees van Deemter, Ielka van der Sluis, and Albert Gatt. 2006. Building a semantically transparent corpus for the generation of referring expressions. In Proceedings of the 4th International Conference on Natural Language Generation (Special Session on Data Sharing and Evaluation), INLG-06. Jette Viethen and Robert Dale. 2008. The use of spatial relations in referring expressions. In Proceedings of the 5th International Conference on Natural Language Generation. 54
6 0.090458088 149 acl-2010-Incorporating Extra-Linguistic Information into Reference Resolution in Collaborative Task Dialogue
7 0.089220092 187 acl-2010-Optimising Information Presentation for Spoken Dialogue Systems
8 0.086894266 167 acl-2010-Learning to Adapt to Unknown Users: Referring Expression Generation in Spoken Dialogue Systems
9 0.077953912 239 acl-2010-Towards Relational POMDPs for Adaptive Dialogue Management
10 0.064467087 199 acl-2010-Preferences versus Adaptation during Referring Expression Generation
11 0.050232518 47 acl-2010-Beetle II: A System for Tutoring and Computational Linguistics Experimentation
12 0.049941007 179 acl-2010-Now, Where Was I? Resumption Strategies for an In-Vehicle Dialogue System
13 0.04973809 194 acl-2010-Phrase-Based Statistical Language Generation Using Graphical Models and Active Learning
14 0.047412809 2 acl-2010-"Was It Good? It Was Provocative." Learning the Meaning of Scalar Adjectives
15 0.046959165 141 acl-2010-Identifying Text Polarity Using Random Walks
16 0.04610284 6 acl-2010-A Game-Theoretic Model of Metaphorical Bargaining
17 0.044385523 142 acl-2010-Importance-Driven Turn-Bidding for Spoken Dialogue Systems
18 0.04376591 46 acl-2010-Bayesian Synchronous Tree-Substitution Grammar Induction and Its Application to Sentence Compression
19 0.043412369 76 acl-2010-Creating Robust Supervised Classifiers via Web-Scale N-Gram Data
20 0.043381121 224 acl-2010-Talking NPCs in a Virtual Game World
topicId topicWeight
[(0, -0.129), (1, 0.054), (2, -0.032), (3, -0.121), (4, -0.049), (5, -0.11), (6, -0.048), (7, 0.064), (8, 0.002), (9, -0.003), (10, 0.011), (11, -0.023), (12, 0.011), (13, 0.029), (14, 0.016), (15, -0.065), (16, 0.142), (17, 0.062), (18, 0.01), (19, 0.092), (20, 0.071), (21, -0.055), (22, -0.022), (23, 0.075), (24, 0.053), (25, -0.048), (26, -0.035), (27, 0.045), (28, -0.129), (29, 0.016), (30, 0.056), (31, -0.233), (32, 0.025), (33, -0.039), (34, 0.032), (35, 0.113), (36, 0.042), (37, 0.147), (38, -0.085), (39, -0.03), (40, 0.03), (41, -0.058), (42, 0.001), (43, 0.039), (44, -0.018), (45, -0.013), (46, -0.05), (47, -0.058), (48, 0.086), (49, -0.077)]
simIndex simValue paperId paperTitle
same-paper 1 0.94425225 35 acl-2010-Automated Planning for Situated Natural Language Generation
Author: Konstantina Garoufi ; Alexander Koller
Abstract: We present a natural language generation approach which models, exploits, and manipulates the non-linguistic context in situated communication, using techniques from AI planning. We show how to generate instructions which deliberately guide the hearer to a location that is convenient for the generation of simple referring expressions, and how to generate referring expressions with context-dependent adjectives. We implement and evaluate our approach in the framework of the Challenge on Generating Instructions in Virtual Environments, finding that it performs well even under the constraints of realtime generation.
2 0.84176403 168 acl-2010-Learning to Follow Navigational Directions
Author: Adam Vogel ; Dan Jurafsky
Abstract: We present a system that learns to follow navigational natural language directions. Where traditional models learn from linguistic annotation or word distributions, our approach is grounded in the world, learning by apprenticeship from routes through a map paired with English descriptions. Lacking an explicit alignment between the text and the reference path makes it difficult to determine what portions of the language describe which aspects of the route. We learn this correspondence with a reinforcement learning algorithm, using the deviation of the route we follow from the intended path as a reward signal. We demonstrate that our system successfully grounds the meaning of spatial terms like above and south into geometric properties of paths.
3 0.7982977 202 acl-2010-Reading between the Lines: Learning to Map High-Level Instructions to Commands
Author: S.R.K. Branavan ; Luke Zettlemoyer ; Regina Barzilay
Abstract: In this paper, we address the task of mapping high-level instructions to sequences of commands in an external environment. Processing these instructions is challenging—they posit goals to be achieved without specifying the steps required to complete them. We describe a method that fills in missing information using an automatically derived environment model that encodes states, transitions, and commands that cause these transitions to happen. We present an efficient approximate approach for learning this environment model as part of a policygradient reinforcement learning algorithm for text interpretation. This design enables learning for mapping high-level instructions, which previous statistical methods cannot handle.1
4 0.43452373 239 acl-2010-Towards Relational POMDPs for Adaptive Dialogue Management
Author: Pierre Lison
Abstract: Open-ended spoken interactions are typically characterised by both structural complexity and high levels of uncertainty, making dialogue management in such settings a particularly challenging problem. Traditional approaches have focused on providing theoretical accounts for either the uncertainty or the complexity of spoken dialogue, but rarely considered the two issues simultaneously. This paper describes ongoing work on a new approach to dialogue management which attempts to fill this gap. We represent the interaction as a Partially Observable Markov Decision Process (POMDP) over a rich state space incorporating both dialogue, user, and environment models. The tractability of the resulting POMDP can be preserved using a mechanism for dynamically constraining the action space based on prior knowledge over locally relevant dialogue structures. These constraints are encoded in a small set of general rules expressed as a Markov Logic network. The first-order expressivity of Markov Logic enables us to leverage the rich relational structure of the problem and efficiently abstract over large regions ofthe state and action spaces.
5 0.42899781 167 acl-2010-Learning to Adapt to Unknown Users: Referring Expression Generation in Spoken Dialogue Systems
Author: Srinivasan Janarthanam ; Oliver Lemon
Abstract: We present a data-driven approach to learn user-adaptive referring expression generation (REG) policies for spoken dialogue systems. Referring expressions can be difficult to understand in technical domains where users may not know the technical ‘jargon’ names of the domain entities. In such cases, dialogue systems must be able to model the user’s (lexical) domain knowledge and use appropriate referring expressions. We present a reinforcement learning (RL) framework in which the sys- tem learns REG policies which can adapt to unknown users online. Furthermore, unlike supervised learning methods which require a large corpus of expert adaptive behaviour to train on, we show that effective adaptive policies can be learned from a small dialogue corpus of non-adaptive human-machine interaction, by using a RL framework and a statistical user simulation. We show that in comparison to adaptive hand-coded baseline policies, the learned policy performs significantly better, with an 18.6% average increase in adaptation accuracy. The best learned policy also takes less dialogue time (average 1.07 min less) than the best hand-coded policy. This is because the learned policies can adapt online to changing evidence about the user’s domain expertise.
6 0.42863798 149 acl-2010-Incorporating Extra-Linguistic Information into Reference Resolution in Collaborative Task Dialogue
7 0.40318394 92 acl-2010-Don't 'Have a Clue'? Unsupervised Co-Learning of Downward-Entailing Operators.
8 0.40119588 231 acl-2010-The Prevalence of Descriptive Referring Expressions in News and Narrative
9 0.38568985 190 acl-2010-P10-5005 k2opt.pdf
10 0.35513562 85 acl-2010-Detecting Experiences from Weblogs
11 0.35088891 187 acl-2010-Optimising Information Presentation for Spoken Dialogue Systems
12 0.35006991 55 acl-2010-Bootstrapping Semantic Analyzers from Non-Contradictory Texts
13 0.31219196 142 acl-2010-Importance-Driven Turn-Bidding for Spoken Dialogue Systems
14 0.29605737 67 acl-2010-Computing Weakest Readings
15 0.2918258 108 acl-2010-Expanding Verb Coverage in Cyc with VerbNet
16 0.28885806 199 acl-2010-Preferences versus Adaptation during Referring Expression Generation
17 0.28648478 182 acl-2010-On the Computational Complexity of Dominance Links in Grammatical Formalisms
18 0.27866131 2 acl-2010-"Was It Good? It Was Provocative." Learning the Meaning of Scalar Adjectives
19 0.27614054 19 acl-2010-A Taxonomy, Dataset, and Classifier for Automatic Noun Compound Interpretation
20 0.27369434 13 acl-2010-A Rational Model of Eye Movement Control in Reading
topicId topicWeight
[(5, 0.325), (14, 0.041), (16, 0.01), (25, 0.058), (36, 0.02), (39, 0.028), (42, 0.043), (59, 0.062), (73, 0.049), (78, 0.054), (83, 0.052), (84, 0.041), (98, 0.089)]
simIndex simValue paperId paperTitle
same-paper 1 0.75535142 35 acl-2010-Automated Planning for Situated Natural Language Generation
Author: Konstantina Garoufi ; Alexander Koller
Abstract: We present a natural language generation approach which models, exploits, and manipulates the non-linguistic context in situated communication, using techniques from AI planning. We show how to generate instructions which deliberately guide the hearer to a location that is convenient for the generation of simple referring expressions, and how to generate referring expressions with context-dependent adjectives. We implement and evaluate our approach in the framework of the Challenge on Generating Instructions in Virtual Environments, finding that it performs well even under the constraints of realtime generation.
2 0.6519776 135 acl-2010-Hindi-to-Urdu Machine Translation through Transliteration
Author: Nadir Durrani ; Hassan Sajjad ; Alexander Fraser ; Helmut Schmid
Abstract: We present a novel approach to integrate transliteration into Hindi-to-Urdu statistical machine translation. We propose two probabilistic models, based on conditional and joint probability formulations, that are novel solutions to the problem. Our models consider both transliteration and translation when translating a particular Hindi word given the context whereas in previous work transliteration is only used for translating OOV (out-of-vocabulary) words. We use transliteration as a tool for disambiguation of Hindi homonyms which can be both translated or transliterated or transliterated differently based on different contexts. We obtain final BLEU scores of 19.35 (conditional prob- ability model) and 19.00 (joint probability model) as compared to 14.30 for a baseline phrase-based system and 16.25 for a system which transliterates OOV words in the baseline system. This indicates that transliteration is useful for more than only translating OOV words for language pairs like Hindi-Urdu.
3 0.56471133 5 acl-2010-A Framework for Figurative Language Detection Based on Sense Differentiation
Author: Daria Bogdanova
Abstract: Various text mining algorithms require the process offeature selection. High-level semantically rich features, such as figurative language uses, speech errors etc., are very promising for such problems as e.g. writing style detection, but automatic extraction of such features is a big challenge. In this paper, we propose a framework for figurative language use detection. This framework is based on the idea of sense differentiation. We describe two algorithms illustrating the mentioned idea. We show then how these algorithms work by applying them to Russian language data.
4 0.42313927 214 acl-2010-Sparsity in Dependency Grammar Induction
Author: Jennifer Gillenwater ; Kuzman Ganchev ; Joao Graca ; Fernando Pereira ; Ben Taskar
Abstract: A strong inductive bias is essential in unsupervised grammar induction. We explore a particular sparsity bias in dependency grammars that encourages a small number of unique dependency types. Specifically, we investigate sparsity-inducing penalties on the posterior distributions of parent-child POS tag pairs in the posterior regularization (PR) framework of Graça et al. (2007). In ex- periments with 12 languages, we achieve substantial gains over the standard expectation maximization (EM) baseline, with average improvement in attachment accuracy of 6.3%. Further, our method outperforms models based on a standard Bayesian sparsity-inducing prior by an average of 4.9%. On English in particular, we show that our approach improves on several other state-of-the-art techniques.
5 0.41706544 70 acl-2010-Contextualizing Semantic Representations Using Syntactically Enriched Vector Models
Author: Stefan Thater ; Hagen Furstenau ; Manfred Pinkal
Abstract: We present a syntactically enriched vector model that supports the computation of contextualized semantic representations in a quasi compositional fashion. It employs a systematic combination of first- and second-order context vectors. We apply our model to two different tasks and show that (i) it substantially outperforms previous work on a paraphrase ranking task, and (ii) achieves promising results on a wordsense similarity task; to our knowledge, it is the first time that an unsupervised method has been applied to this task.
6 0.41685462 71 acl-2010-Convolution Kernel over Packed Parse Forest
7 0.41638741 158 acl-2010-Latent Variable Models of Selectional Preference
8 0.41565078 65 acl-2010-Complexity Metrics in an Incremental Right-Corner Parser
9 0.41455895 172 acl-2010-Minimized Models and Grammar-Informed Initialization for Supertagging with Highly Ambiguous Lexicons
10 0.41367662 17 acl-2010-A Structured Model for Joint Learning of Argument Roles and Predicate Senses
11 0.41361779 21 acl-2010-A Tree Transducer Model for Synchronous Tree-Adjoining Grammars
12 0.413524 130 acl-2010-Hard Constraints for Grammatical Function Labelling
13 0.41341433 120 acl-2010-Fully Unsupervised Core-Adjunct Argument Classification
14 0.41316241 168 acl-2010-Learning to Follow Navigational Directions
15 0.41297388 248 acl-2010-Unsupervised Ontology Induction from Text
16 0.41275549 231 acl-2010-The Prevalence of Descriptive Referring Expressions in News and Narrative
17 0.41098952 211 acl-2010-Simple, Accurate Parsing with an All-Fragments Grammar
18 0.41026169 162 acl-2010-Learning Common Grammar from Multilingual Corpus
19 0.40888369 251 acl-2010-Using Anaphora Resolution to Improve Opinion Target Identification in Movie Reviews
20 0.40865457 55 acl-2010-Bootstrapping Semantic Analyzers from Non-Contradictory Texts