acl acl2011 acl2011-308 knowledge-graph by maker-knowledge-mining

308 acl-2011-Towards a Framework for Abstractive Summarization of Multimodal Documents


Source: pdf

Author: Charles Greenbacker

Abstract: We propose a framework for generating an abstractive summary from a semantic model of a multimodal document. We discuss the type of model required, the means by which it can be constructed, how the content of the model is rated and selected, and the method of realizing novel sentences for the summary. To this end, we introduce a metric called information density used for gauging the importance of content obtained from text and graphical sources.

Reference: text


Summary: the most important sentenses genereted by tfidf model

sentIndex sentText sentNum sentScore

1 of Computer & Information Sciences University of Delaware Newark, Delaware, USA charl ieg@ ci s . [sent-3, score-0.149]

2 edu Abstract We propose a framework for generating an abstractive summary from a semantic model of a multimodal document. [sent-5, score-0.775]

3 We discuss the type of model required, the means by which it can be constructed, how the content of the model is rated and selected, and the method of realizing novel sentences for the summary. [sent-6, score-0.25]

4 To this end, we introduce a metric called information density used for gauging the importance of content obtained from text and graphical sources. [sent-7, score-0.314]

5 1 Introduction The automatic summarization of text is a prominent task in the field of natural language processing (NLP). [sent-8, score-0.154]

6 While significant achievements have been made using statistical analysis and sentence extraction, “true abstractive summarization remains a researcher’s dream” (Radev et al. [sent-9, score-0.489]

7 Although existing systems produce high-quality summaries of relatively simple articles, there are limitations as to the types of documents these systems can handle. [sent-11, score-0.296]

8 One such limitation is the summarization of multimodal documents: no existing system is able to incorporate the non-text portions of a document (e. [sent-12, score-0.503]

9 (2006) showed that the content of information graphics is often not repeated in the article’s text, meaning important information may be overlooked if the graphical content is not included in the summary. [sent-16, score-0.589]

10 Systems that perform statistical analysis of text and extract sentences from the original article to assemble a summary cannot access the information contained in non-text components, 75 let alone seamlessly combine that information with the extracted text. [sent-17, score-0.38]

11 The problem is that information from the text and graphical components can only be integrated at the conceptual level, necessitating a semantic understanding of the underlying concepts. [sent-18, score-0.506]

12 Our proposed framework enables the generation of abstractive summaries from unified semantic models, regardless of the original format of the in- formation sources. [sent-19, score-0.682]

13 We contend that this framework is more akin to the human process of conceptual integration and regeneration in writing an abstract, as compared to the traditional NLP techniques of rating and extracting sentences to form a summary. [sent-20, score-0.479]

14 Most work in this area has focused on extractive summarization, assembling the summary from sentences representing the information in a document (Kupiec et al. [sent-23, score-0.193]

15 Discourse structure (Marcu, 1997) also helps indicate the most important sentences. [sent-26, score-0.044]

16 However, a few efforts have been directed towards abstractive summaries, including the modification (i. [sent-32, score-0.371]

17 , editing and rewriting) of extracted sentences (Jing and McKeown, 1999) and the generation of novel sentences based on a deeper understanding of the concepts being described. [sent-34, score-0.175]

18 Lexical chains, which capture relationships between related terms in a document, have shown promise as an intermediate representation for producing summaries (Barzilay and Elhadad, 1997). [sent-35, score-0.215]

19 Our work shares similarities with the knowledge-based text condensation model of Reimer and Hahn (1988), as well as with Rau et al. [sent-36, score-0.06]

20 (1989), who developed an information extraction approach for conceptual information summarization. [sent-37, score-0.175]

21 While we also build a conceptual model, we believe our method of construction will produce a richer representation. [sent-38, score-0.175]

22 Moreover, Reimer and Hahn did not actually produce a natural language summary, but rather a condensed text graph. [sent-39, score-0.05]

23 Efforts towards the summarization of multimodal documents have included na¨ ıve approaches relying on image captions and direct references to the image in the text (Bhatia et al. [sent-40, score-0.995]

24 , 2009), while contentbased image analysis and NLP techniques are being combined for multimodal document indexing and retrieval in the medical domain (N´ ev´ eol et al. [sent-41, score-0.572]

25 3 Method Our method consists of the following steps: building the semantic model, rating the informational content, and generating a summary. [sent-43, score-0.234]

26 We construct the semantic model in a knowledge representation based on typed, structured objects organized under a foundational ontology (McDonald, 2000). [sent-44, score-0.242]

27 To analyze the text, we use Sparser,1 a linguistically-sound, phrase structure-based chart parser with an extensive and extendible semantic grammar (McDonald, 1992). [sent-45, score-0.283]

28 For the purposes of this proposal, we assume a relatively complete semantic grammar exists for the domain of documents to be summarized. [sent-46, score-0.235]

29 In the prototype implementation (currently in progress), we are manually extending an existing grammar on an asneeded basis, with plans for large-scale learning of new rules and ontology definitions as future work. [sent-47, score-0.444]

30 Projects like the Never-Ending Language Learner 1https : / / github . [sent-48, score-0.054]

31 Although our framework is general enough to cover any image type, as well as other modalities (e. [sent-51, score-0.372]

32 , audio, video), since image understanding research has not yet developed tools capable of extracting semantic content from every possible image, we must restrict our focus to a limited class of images for the prototype implementation. [sent-53, score-0.736]

33 Information graphics, such as bar charts and line graphs, are commonly found in popular media (e. [sent-54, score-0.212]

34 To integrate this graphical content, we use the SIGHT system (Demir et al. [sent-57, score-0.126]

35 , 2010b) which identifies the intended message of a bar chart or line graph along with other salient propositions conveyed by the graphic. [sent-58, score-0.397]

36 Extending the prototype to incorporate other modalities would not entail a significant change to the framework. [sent-59, score-0.293]

37 However, it would require adding a module capable of mapping the particular modality to its un- derlying message-level semantic content. [sent-60, score-0.271]

38 The next sections provide detail regarding the steps of our method, which will be illustrated on a short article from the May 29, 2006 edition of Businessweek magazine entitled, “Will Medtronic’s Pulse Quicken? [sent-61, score-0.24]

39 ”2 This particular article was chosen due to good coverage in the existing Sparser grammar for the business news domain, and because it appears in the corpus of multimodal documents made available by the SIGHT project. [sent-62, score-0.57]

40 1 Semantic Modeling Figure 1 shows a high-level (low-detail) overview of the type of semantic model we can build using Sparser and SIGHT. [sent-64, score-0.092]

41 This particular example models the article text (including title) and line graph from the Medtronic article. [sent-65, score-0.193]

42 Each box represents an individual concept recognized in the document. [sent-66, score-0.116]

43 In the interest of space, the individual attributes of the model entries have been omitted from this diagram, but are available in Figure 2, which zooms into a fragment of the model showing the concepts that are eventually rated most salient (Section 3. [sent-68, score-0.348]

44 com/ maga z ine / content / 0 6_2 2 /b3 9 8 6 12 0 . [sent-71, score-0.21]

45 Figure 1: High-level overview of semantic model for Medtronic article. [sent-73, score-0.092]


similar papers computed by tfidf model

tfidf for this paper:

wordName wordTfidf (topN-words)

[('abstractive', 0.269), ('multimodal', 0.234), ('medtronic', 0.223), ('image', 0.206), ('conceptual', 0.175), ('summaries', 0.167), ('graphics', 0.162), ('summarization', 0.154), ('charl', 0.149), ('greenbacker', 0.149), ('article', 0.144), ('sparser', 0.135), ('reimer', 0.131), ('prototype', 0.128), ('summary', 0.127), ('graphical', 0.126), ('content', 0.122), ('delaware', 0.121), ('sight', 0.113), ('hahn', 0.113), ('bar', 0.113), ('modalities', 0.113), ('semantic', 0.092), ('ontology', 0.09), ('ine', 0.088), ('documents', 0.08), ('concepts', 0.078), ('rated', 0.078), ('rating', 0.076), ('mcdonald', 0.075), ('images', 0.07), ('box', 0.068), ('document', 0.066), ('magazines', 0.066), ('pulse', 0.066), ('extendible', 0.066), ('gauging', 0.066), ('informational', 0.066), ('captions', 0.066), ('dream', 0.066), ('eol', 0.066), ('witbrock', 0.066), ('contend', 0.066), ('zooms', 0.066), ('achievements', 0.066), ('derlying', 0.066), ('newark', 0.066), ('capable', 0.065), ('salient', 0.063), ('attributes', 0.063), ('grammar', 0.063), ('chart', 0.062), ('extending', 0.062), ('portion', 0.061), ('condensation', 0.06), ('rau', 0.06), ('accompanying', 0.06), ('bus', 0.06), ('propositions', 0.06), ('necessitating', 0.06), ('foundational', 0.06), ('kupiec', 0.06), ('overlooked', 0.057), ('regeneration', 0.057), ('assemble', 0.057), ('enables', 0.055), ('github', 0.054), ('efforts', 0.053), ('understanding', 0.053), ('framework', 0.053), ('seamlessly', 0.052), ('akin', 0.052), ('mittal', 0.052), ('entail', 0.052), ('magazine', 0.052), ('plans', 0.052), ('aone', 0.052), ('conveyed', 0.05), ('realizing', 0.05), ('charts', 0.05), ('condensed', 0.05), ('entitled', 0.05), ('ev', 0.05), ('angle', 0.05), ('line', 0.049), ('towards', 0.049), ('existing', 0.049), ('modality', 0.048), ('boxes', 0.048), ('formats', 0.048), ('relationships', 0.048), ('concept', 0.048), ('rewriting', 0.046), ('formation', 0.046), ('elhadad', 0.046), ('researcher', 0.045), ('helps', 0.044), ('edition', 0.044), ('newspapers', 0.044), ('editing', 0.044)]

similar papers list:

simIndex simValue paperId paperTitle

same-paper 1 1.0000002 308 acl-2011-Towards a Framework for Abstractive Summarization of Multimodal Documents

Author: Charles Greenbacker

Abstract: We propose a framework for generating an abstractive summary from a semantic model of a multimodal document. We discuss the type of model required, the means by which it can be constructed, how the content of the model is rated and selected, and the method of realizing novel sentences for the summary. To this end, we introduce a metric called information density used for gauging the importance of content obtained from text and graphical sources.

2 0.22058949 21 acl-2011-A Pilot Study of Opinion Summarization in Conversations

Author: Dong Wang ; Yang Liu

Abstract: This paper presents a pilot study of opinion summarization on conversations. We create a corpus containing extractive and abstractive summaries of speaker’s opinion towards a given topic using 88 telephone conversations. We adopt two methods to perform extractive summarization. The first one is a sentence-ranking method that linearly combines scores measured from different aspects including topic relevance, subjectivity, and sentence importance. The second one is a graph-based method, which incorporates topic and sentiment information, as well as additional information about sentence-to-sentence relations extracted based on dialogue structure. Our evaluation results show that both methods significantly outperform the baseline approach that extracts the longest utterances. In particular, we find that incorporating dialogue structure in the graph-based method contributes to the improved system performance.

3 0.14119765 270 acl-2011-SciSumm: A Multi-Document Summarization System for Scientific Articles

Author: Nitin Agarwal ; Ravi Shankar Reddy ; Kiran GVR ; Carolyn Penstein Rose

Abstract: In this demo, we present SciSumm, an interactive multi-document summarization system for scientific articles. The document collection to be summarized is a list of papers cited together within the same source article, otherwise known as a co-citation. At the heart of the approach is a topic based clustering of fragments extracted from each article based on queries generated from the context surrounding the co-cited list of papers. This analysis enables the generation of an overview of common themes from the co-cited papers that relate to the context in which the co-citation was found. SciSumm is currently built over the 2008 ACL Anthology, however the gen- eralizable nature of the summarization techniques and the extensible architecture makes it possible to use the system with other corpora where a citation network is available. Evaluation results on the same corpus demonstrate that our system performs better than an existing widely used multi-document summarization system (MEAD).

4 0.12890415 76 acl-2011-Comparative News Summarization Using Linear Programming

Author: Xiaojiang Huang ; Xiaojun Wan ; Jianguo Xiao

Abstract: Comparative News Summarization aims to highlight the commonalities and differences between two comparable news topics. In this study, we propose a novel approach to generating comparative news summaries. We formulate the task as an optimization problem of selecting proper sentences to maximize the comparativeness within the summary and the representativeness to both news topics. We consider semantic-related cross-topic concept pairs as comparative evidences, and consider topic-related concepts as representative evidences. The optimization problem is addressed by using a linear programming model. The experimental results demonstrate the effectiveness of our proposed model.

5 0.12822582 98 acl-2011-Discovery of Topically Coherent Sentences for Extractive Summarization

Author: Asli Celikyilmaz ; Dilek Hakkani-Tur

Abstract: Extractive methods for multi-document summarization are mainly governed by information overlap, coherence, and content constraints. We present an unsupervised probabilistic approach to model the hidden abstract concepts across documents as well as the correlation between these concepts, to generate topically coherent and non-redundant summaries. Based on human evaluations our models generate summaries with higher linguistic quality in terms of coherence, readability, and redundancy compared to benchmark systems. Although our system is unsupervised and optimized for topical coherence, we achieve a 44.1 ROUGE on the DUC-07 test set, roughly in the range of state-of-the-art supervised models.

6 0.12816215 251 acl-2011-Probabilistic Document Modeling for Syntax Removal in Text Summarization

7 0.12329997 326 acl-2011-Using Bilingual Information for Cross-Language Document Summarization

8 0.11896233 187 acl-2011-Jointly Learning to Extract and Compress

9 0.10717308 280 acl-2011-Sentence Ordering Driven by Local and Global Coherence for Summary Generation

10 0.10201069 71 acl-2011-Coherent Citation-Based Summarization of Scientific Papers

11 0.10191506 18 acl-2011-A Latent Topic Extracting Method based on Events in a Document and its Application

12 0.098283909 47 acl-2011-Automatic Assessment of Coverage Quality in Intelligence Reports

13 0.095627718 201 acl-2011-Learning From Collective Human Behavior to Introduce Diversity in Lexical Choice

14 0.088616431 99 acl-2011-Discrete vs. Continuous Rating Scales for Language Evaluation in NLP

15 0.079851337 255 acl-2011-Query Snowball: A Co-occurrence-based Approach to Multi-document Summarization for Question Answering

16 0.073704109 4 acl-2011-A Class of Submodular Functions for Document Summarization

17 0.065729953 14 acl-2011-A Hierarchical Model of Web Summaries

18 0.065665014 288 acl-2011-Subjective Natural Language Problems: Motivations, Applications, Characterizations, and Implications

19 0.063893497 156 acl-2011-IMASS: An Intelligent Microblog Analysis and Summarization System

20 0.061351143 53 acl-2011-Automatically Evaluating Text Coherence Using Discourse Relations


similar papers computed by lsi model

lsi for this paper:

topicId topicWeight

[(0, 0.177), (1, 0.086), (2, -0.08), (3, 0.069), (4, -0.062), (5, 0.012), (6, -0.082), (7, 0.167), (8, 0.011), (9, -0.07), (10, -0.109), (11, -0.021), (12, -0.115), (13, 0.006), (14, -0.156), (15, -0.021), (16, 0.078), (17, 0.043), (18, 0.02), (19, 0.061), (20, -0.019), (21, -0.02), (22, 0.049), (23, -0.015), (24, -0.018), (25, -0.037), (26, 0.057), (27, -0.097), (28, 0.062), (29, -0.017), (30, -0.011), (31, 0.065), (32, 0.03), (33, -0.084), (34, 0.051), (35, 0.02), (36, 0.037), (37, 0.02), (38, -0.002), (39, 0.095), (40, -0.049), (41, 0.057), (42, 0.022), (43, -0.029), (44, -0.022), (45, -0.031), (46, -0.009), (47, -0.01), (48, 0.021), (49, 0.076)]

similar papers list:

simIndex simValue paperId paperTitle

same-paper 1 0.95109528 308 acl-2011-Towards a Framework for Abstractive Summarization of Multimodal Documents

Author: Charles Greenbacker

Abstract: We propose a framework for generating an abstractive summary from a semantic model of a multimodal document. We discuss the type of model required, the means by which it can be constructed, how the content of the model is rated and selected, and the method of realizing novel sentences for the summary. To this end, we introduce a metric called information density used for gauging the importance of content obtained from text and graphical sources.

2 0.78338599 270 acl-2011-SciSumm: A Multi-Document Summarization System for Scientific Articles

Author: Nitin Agarwal ; Ravi Shankar Reddy ; Kiran GVR ; Carolyn Penstein Rose

Abstract: In this demo, we present SciSumm, an interactive multi-document summarization system for scientific articles. The document collection to be summarized is a list of papers cited together within the same source article, otherwise known as a co-citation. At the heart of the approach is a topic based clustering of fragments extracted from each article based on queries generated from the context surrounding the co-cited list of papers. This analysis enables the generation of an overview of common themes from the co-cited papers that relate to the context in which the co-citation was found. SciSumm is currently built over the 2008 ACL Anthology, however the gen- eralizable nature of the summarization techniques and the extensible architecture makes it possible to use the system with other corpora where a citation network is available. Evaluation results on the same corpus demonstrate that our system performs better than an existing widely used multi-document summarization system (MEAD).

3 0.75294495 76 acl-2011-Comparative News Summarization Using Linear Programming

Author: Xiaojiang Huang ; Xiaojun Wan ; Jianguo Xiao

Abstract: Comparative News Summarization aims to highlight the commonalities and differences between two comparable news topics. In this study, we propose a novel approach to generating comparative news summaries. We formulate the task as an optimization problem of selecting proper sentences to maximize the comparativeness within the summary and the representativeness to both news topics. We consider semantic-related cross-topic concept pairs as comparative evidences, and consider topic-related concepts as representative evidences. The optimization problem is addressed by using a linear programming model. The experimental results demonstrate the effectiveness of our proposed model.

4 0.74612606 187 acl-2011-Jointly Learning to Extract and Compress

Author: Taylor Berg-Kirkpatrick ; Dan Gillick ; Dan Klein

Abstract: We learn a joint model of sentence extraction and compression for multi-document summarization. Our model scores candidate summaries according to a combined linear model whose features factor over (1) the n-gram types in the summary and (2) the compressions used. We train the model using a marginbased objective whose loss captures end summary quality. Because of the exponentially large set of candidate summaries, we use a cutting-plane algorithm to incrementally detect and add active constraints efficiently. Inference in our model can be cast as an ILP and thereby solved in reasonable time; we also present a fast approximation scheme which achieves similar performance. Our jointly extracted and compressed summaries outperform both unlearned baselines and our learned extraction-only system on both ROUGE and Pyramid, without a drop in judged linguistic quality. We achieve the highest published ROUGE results to date on the TAC 2008 data set.

5 0.72197729 201 acl-2011-Learning From Collective Human Behavior to Introduce Diversity in Lexical Choice

Author: Vahed Qazvinian ; Dragomir R. Radev

Abstract: We analyze collective discourse, a collective human behavior in content generation, and show that it exhibits diversity, a property of general collective systems. Using extensive analysis, we propose a novel paradigm for designing summary generation systems that reflect the diversity of perspectives seen in reallife collective summarization. We analyze 50 sets of summaries written by human about the same story or artifact and investigate the diversity of perspectives across these summaries. We show how different summaries use various phrasal information units (i.e., nuggets) to express the same atomic semantic units, called factoids. Finally, we present a ranker that employs distributional similarities to build a net- work of words, and captures the diversity of perspectives by detecting communities in this network. Our experiments show how our system outperforms a wide range of other document ranking systems that leverage diversity.

6 0.71890914 47 acl-2011-Automatic Assessment of Coverage Quality in Intelligence Reports

7 0.70162666 326 acl-2011-Using Bilingual Information for Cross-Language Document Summarization

8 0.69044799 21 acl-2011-A Pilot Study of Opinion Summarization in Conversations

9 0.68889868 251 acl-2011-Probabilistic Document Modeling for Syntax Removal in Text Summarization

10 0.66254759 80 acl-2011-ConsentCanvas: Automatic Texturing for Improved Readability in End-User License Agreements

11 0.63862997 4 acl-2011-A Class of Submodular Functions for Document Summarization

12 0.62743402 255 acl-2011-Query Snowball: A Co-occurrence-based Approach to Multi-document Summarization for Question Answering

13 0.60891747 98 acl-2011-Discovery of Topically Coherent Sentences for Extractive Summarization

14 0.52479911 280 acl-2011-Sentence Ordering Driven by Local and Global Coherence for Summary Generation

15 0.51670325 71 acl-2011-Coherent Citation-Based Summarization of Scientific Papers

16 0.51450944 156 acl-2011-IMASS: An Intelligent Microblog Analysis and Summarization System

17 0.51305962 67 acl-2011-Clairlib: A Toolkit for Natural Language Processing, Information Retrieval, and Network Analysis

18 0.5076822 291 acl-2011-SystemT: A Declarative Information Extraction System

19 0.48927703 51 acl-2011-Automatic Headline Generation using Character Cross-Correlation

20 0.47004765 18 acl-2011-A Latent Topic Extracting Method based on Events in a Document and its Application


similar papers computed by lda model

lda for this paper:

topicId topicWeight

[(1, 0.013), (5, 0.037), (16, 0.011), (17, 0.045), (26, 0.025), (31, 0.13), (35, 0.16), (37, 0.067), (39, 0.039), (41, 0.057), (55, 0.022), (59, 0.052), (72, 0.025), (91, 0.034), (96, 0.201)]

similar papers list:

simIndex simValue paperId paperTitle

same-paper 1 0.86943996 308 acl-2011-Towards a Framework for Abstractive Summarization of Multimodal Documents

Author: Charles Greenbacker

Abstract: We propose a framework for generating an abstractive summary from a semantic model of a multimodal document. We discuss the type of model required, the means by which it can be constructed, how the content of the model is rated and selected, and the method of realizing novel sentences for the summary. To this end, we introduce a metric called information density used for gauging the importance of content obtained from text and graphical sources.

2 0.84639251 218 acl-2011-MemeTube: A Sentiment-based Audiovisual System for Analyzing and Displaying Microblog Messages

Author: Cheng-Te Li ; Chien-Yuan Wang ; Chien-Lin Tseng ; Shou-De Lin

Abstract: Micro-blogging services provide platforms for users to share their feelings and ideas on the move. In this paper, we present a search-based demonstration system, called MemeTube, to summarize the sentiments of microblog messages in an audiovisual manner. MemeTube provides three main functions: (1) recognizing the sentiments of messages (2) generating music melody automatically based on detected sentiments, and (3) produce an animation of real-time piano playing for audiovisual display. Our MemeTube system can be accessed via: http://mslab.csie.ntu.edu.tw/memetube/ .

3 0.84016144 317 acl-2011-Underspecifying and Predicting Voice for Surface Realisation Ranking

Author: Sina Zarriess ; Aoife Cahill ; Jonas Kuhn

Abstract: This paper addresses a data-driven surface realisation model based on a large-scale reversible grammar of German. We investigate the relationship between the surface realisation performance and the character of the input to generation, i.e. its degree of underspecification. We extend a syntactic surface realisation system, which can be trained to choose among word order variants, such that the candidate set includes active and passive variants. This allows us to study the interaction of voice and word order alternations in realistic German corpus data. We show that with an appropriately underspecified input, a linguistically informed realisation model trained to regenerate strings from the underlying semantic representation achieves 91.5% accuracy (over a baseline of 82.5%) in the prediction of the original voice. 1

4 0.83608973 75 acl-2011-Combining Morpheme-based Machine Translation with Post-processing Morpheme Prediction

Author: Ann Clifton ; Anoop Sarkar

Abstract: This paper extends the training and tuning regime for phrase-based statistical machine translation to obtain fluent translations into morphologically complex languages (we build an English to Finnish translation system) . Our methods use unsupervised morphology induction. Unlike previous work we focus on morphologically productive phrase pairs – our decoder can combine morphemes across phrase boundaries. Morphemes in the target language may not have a corresponding morpheme or word in the source language. Therefore, we propose a novel combination of post-processing morphology prediction with morpheme-based translation. We show, using both automatic evaluation scores and linguistically motivated analyses of the output, that our methods outperform previously proposed ones and pro- vide the best known results on the EnglishFinnish Europarl translation task. Our methods are mostly language independent, so they should improve translation into other target languages with complex morphology. 1 Translation and Morphology Languages with rich morphological systems present significant hurdles for statistical machine translation (SMT) , most notably data sparsity, source-target asymmetry, and problems with automatic evaluation. In this work, we propose to address the problem of morphological complexity in an Englishto-Finnish MT task within a phrase-based translation framework. We focus on unsupervised segmentation methods to derive the morphological information supplied to the MT model in order to provide coverage on very large datasets and for languages with few hand-annotated 32 resources. In fact, in our experiments, unsupervised morphology always outperforms the use of a hand-built morphological analyzer. Rather than focusing on a few linguistically motivated aspects of Finnish morphological behaviour, we develop techniques for handling morphological complexity in general. We chose Finnish as our target language for this work, because it exemplifies many of the problems morphologically complex languages present for SMT. Among all the languages in the Europarl data-set, Finnish is the most difficult language to translate from and into, as was demonstrated in the MT Summit shared task (Koehn, 2005) . Another reason is the current lack of knowledge about how to apply SMT successfully to agglutinative languages like Turkish or Finnish. Our main contributions are: 1) the introduction of the notion of segmented translation where we explicitly allow phrase pairs that can end with a dangling morpheme, which can connect with other morphemes as part of the translation process, and 2) the use of a fully segmented translation model in combination with a post-processing morpheme prediction system, using unsupervised morphology induction. Both of these approaches beat the state of the art on the English-Finnish translation task. Morphology can express both content and function categories, and our experiments show that it is important to use morphology both within the translation model (for morphology with content) and outside it (for morphology contributing to fluency) . Automatic evaluation measures for MT, BLEU (Papineni et al., 2002), WER (Word Error Rate) and PER (Position Independent Word Error Rate) use the word as the basic unit rather than morphemes. In a word comProce dPinogrstla ofn tdh,e O 4r9etghon A,n Jnu nael 1 M9-e 2t4i,n2g 0 o1f1 t.he ?c A2s0s1o1ci Aatsiosonc fioartio Cno fmorpu Ctoamtiopnuatalt Lioin gauli Lsitnicgsu,i psatgices 32–42, prised of multiple morphemes, getting even a single morpheme wrong means the entire word is wrong. In addition to standard MT evaluation measures, we perform a detailed linguistic analysis of the output. Our proposed approaches are significantly better than the state of the art, achieving the highest reported BLEU scores on the English-Finnish Europarl version 3 data-set. Our linguistic analysis shows that our models have fewer morpho-syntactic errors compared to the word-based baseline. 2 2.1 Models Baseline Models We set up three baseline models for comparison in this work. The first is a basic wordbased model (called Baseline in the results) ; we trained this on the original unsegmented version of the text. Our second baseline is a factored translation model (Koehn and Hoang, 2007) (called Factored) , which used as factors the word, “stem” 1 and suffix. These are derived from the same unsupervised segmentation model used in other experiments. The results (Table 3) show that a factored model was unable to match the scores of a simple wordbased baseline. We hypothesize that this may be an inherently difficult representational form for a language with the degree of morphological complexity found in Finnish. Because the morphology generation must be precomputed, for languages with a high degree of morphological complexity, the combinatorial explosion makes it unmanageable to capture the full range of morphological productivity. In addition, because the morphological variants are generated on a per-word basis within a given phrase, it excludes productive morphological combination across phrase boundaries and makes it impossible for the model to take into account any longdistance dependencies between morphemes. We conclude from this result that it may be more useful for an agglutinative language to use morphology beyond the confines of the phrasal unit, and condition its generation on more than just the local target stem. In order to compare the 1see Section 2.2. 33 performance of unsupervised segmentation for translation, our third baseline is a segmented translation model based on a supervised segmentation model (called Sup) , using the hand-built Omorfi morphological analyzer (Pirinen and Listenmaa, 2007) , which provided slightly higher BLEU scores than the word-based baseline. 2.2 Segmented Translation For segmented translation models, it cannot be taken for granted that greater linguistic accuracy in segmentation yields improved translation (Chang et al. , 2008) . Rather, the goal in segmentation for translation is instead to maximize the amount of lexical content-carrying morphology, while generalizing over the information not helpful for improving the translation model. We therefore trained several different segmentation models, considering factors of granularity, coverage, and source-target symmetry. We performed unsupervised segmentation of the target data, using Morfessor (Creutz and Lagus, 2005) and Paramor (Monson, 2008) , two top systems from the Morpho Challenge 2008 (their combined output was the Morpho Challenge winner) . However, translation models based upon either Paramor alone or the combined systems output could not match the wordbased baseline, so we concentrated on Morfessor. Morfessor uses minimum description length criteria to train a HMM-based segmentation model. When tested against a human-annotated gold standard of linguistic morpheme segmentations for Finnish, this algorithm outperforms competing unsupervised methods, achieving an F-score of 67.0% on a 3 million sentence corpus (Creutz and Lagus, 2006) . Varying the perplexity threshold in Morfessor does not segment more word types, but rather over-segments the same word types. In order to get robust, common segmentations, we trained the segmenter on the 5000 most frequent words2 ; we then used this to segment the entire data set. In order to improve coverage, we then further segmented 2For the factored model baseline we also used the same setting perplexity = 30, 5,000 most frequent words, but with all but the last suffix collapsed and called the “stem” . TabHMleoat1nr:gplhiMngor phermphocTur631ae04in, 81c9ie03ns67gi,64n0S14e567theTp 2rsa51t, 29Se 3t168able and in translation. any word type that contained a match from the most frequent suffix set, looking for the longest matching suffix character string. We call this method Unsup L-match. After the segmentation, word-internal morpheme boundary markers were inserted into the segmented text to be used to reconstruct the surface forms in the MT output. We then trained the Moses phrase-based system (Koehn et al., 2007) on the segmented and marked text. After decoding, it was a simple matter to join together all adjacent morphemes with word-internal boundary markers to reconstruct the surface forms. Figure 1(a) gives the full model overview for all the variants of the segmented translation model (supervised/unsupervised; with and without the Unsup L-match procedure) . Table 1shows how morphemes are being used in the MT system. Of the phrases that included segmentations (‘Morph’ in Table 1) , roughly a third were ‘productive’, i.e. had a hanging morpheme (with a form such as stem+) that could be joined to a suffix (‘Hanging Morph’ in Table 1) . However, in phrases used while decoding the development and test data, roughly a quarter of the phrases that generated the translated output included segmentations, but of these, only a small fraction (6%) had a hanging morpheme; and while there are many possible reasons to account for this we were unable to find a single convincing cause. 2.3 Morphology Generation Morphology generation as a post-processing step allows major vocabulary reduction in the translation model, and allows the use of morphologically targeted features for modeling inflection. A possible disadvantage of this approach is that in this model there is no opportunity to con34 sider the morphology in translation since it is removed prior to training the translation model. Morphology generation models can use a variety of bilingual and contextual information to capture dependencies between morphemes, often more long-distance than what is possible using n-gram language models over morphemes in the segmented model. Similar to previous work (Minkov et al. , 2007; Toutanova et al. , 2008) , we model morphology generation as a sequence learning problem. Un- like previous work, we use unsupervised morphology induction and use automatically generated suffix classes as tags. The first phase of our morphology prediction model is to train a MT system that produces morphologically simplified word forms in the target language. The output word forms are complex stems (a stem and some suffixes) but still missing some important suffix morphemes. In the second phase, the output of the MT decoder is then tagged with a sequence of abstract suffix tags. In particular, the output of the MT decoder is a sequence of complex stems denoted by x and the output is a sequence of suffix class tags denoted by y. We use a list of parts from (x,y) and map to a d-dimensional feature vector Φ(x, y) , with each dimension being a real number. We infer the best sequence of tags using: F(x) = argymaxp(y | x,w) where F(x) returns the highest scoring output y∗ . A conditional random field (CRF) (Lafferty et al. , 2001) defines the conditional probability as a linear score for each candidate y and a global normalization term: logp(y | x, w) = Φ(x, y) · w − log Z where Z = Py0∈ exp(Φ(x, y0) · w) . We use stochastiPc gradient descent (using crfsgd3) to train the weight vector w. So far, this is all off-the-shelf sequence learning. However, the output y∗ from the CRF decoder is still only a sequence of abstract suffix tags. The third and final phase in our morphology prediction model GEN(x) 3 http://leon. bottou. org/projects/sgd English Training Data words Finnish Training Data words Morphological Pre-Processing stem+ +morph MT System Alignment: word word word stem+ +morph stem stem+ +morph Post-Process: Morph Re-Stitching Fully inflected surface form Evaluation against original reference (a) Segmented Translation Model English Training Data words Finnish Training Data Morphological Pre-Prowceosrdsisng 1 stem+ +morph1+ +morph2 Morphological Pre-Processing 2 stem+ +morph1+ MPosrpthe-mPRr+eo-+cSmetsio crhp1i:nhg+swteomrd+ MA+lTmigwnSomyrspdthen 1mt:+ wsotermd complex stem: stem+morph1+ MPo rpsht-oPlro gcyesGse2n:erCaRtioFnstem+morph1+ morph2sLuarnfagcueagfeorMmomdealp ing Fully inflected surface form Evaluation against original reference (b) Post-Processing Model Translation & Generation Figure 1: Training and testing pipelines for the SMT models. is to take the abstract suffix tag sequence y∗ and then map it into fully inflected word forms, and rank those outputs using a morphemic language model. The abstract suffix tags are extracted from the unsupervised morpheme learning process, and are carefully designed to enable CRF training and decoding. We call this model CRFLM for short. Figure 1(b) shows the full pipeline and Figure 2 shows a worked example of all the steps involved. We use the morphologically segmented training data (obtained using the segmented corpus described in Section 2.24) and remove selected suffixes to create a morphologically simplified version of the training data. The MT model is trained on the morphologically simplified training data. The output from the MT system is then used as input to the CRF model. The CRF model was trained on a ∼210,000 Finnish sentences, consisting noefd d∼ o1n.5 a am ∼il2li1o0n,0 tokens; tishhe 2,000 cseens,te cnoncse Europarl t.e5s tm isl eito nco tnoskiesntesd; hoef 41,434 stem tokens. The labels in the output sequence y were obtained by selecting the most productive 150 stems, and then collapsing certain vowels into equivalence classes corresponding to Finnish vowel harmony patterns. Thus 4Note that unlike Section 2.2 we do not use Unsup L-match because when evaluating the CRF model on the suffix prediction task it obtained 95.61% without using Unsup L-match and 82.99% when using Unsup L-match. 35 variants -k¨ o and -ko become vowel-generic enclitic particle -kO, and variants -ss ¨a and -ssa become the vowel-generic inessive case marker -ssA, etc. This is the only language-specific component of our translation model. However, we expect this approach to work for other agglutinative languages as well. For fusional languages like Spanish, another mapping from suffix to abstract tags might be needed. These suffix transformations to their equivalence classes prevent morphophonemic variants of the same morpheme from competing against each other in the prediction model. This resulted in 44 possible label outputs per stem which was a reasonable sized tag-set for CRF training. The CRF was trained on monolingual features of the segmented text for suffix prediction, where t is the current token: Word Stem st−n, .., st, .., st+n(n = 4) Morph Prediction yt−2 , yt−1 , yt With this simple feature set, we were able to use features over longer distances, resulting in a total of 1,110,075 model features. After CRF based recovery of the suffix tag sequence, we use a bigram language model trained on a full segmented version on the training data to recover the original vowels. We used bigrams only, because the suffix vowel harmony alternation depends only upon the preceding phonemes in the word from which it was segmented. original training koskevaa mietint o¨ ¨a data: k ¨asitell ¨a ¨an segmentation: koske+ +va+ +a mietint ¨o+ + a¨ k a¨si+ +te+ +ll a¨+ + a¨+ +n (train bigram language model with mapping A = { a , a }) map n fi bniaglr asmuff liaxn gtou agbest mraocdte tag-set: koske+ +va+ +A mietint ¨o+ +A k ¨asi+ +te+ +ll ¨a+ + ¨a+ +n (train CRF model to predict the final suffix) peeling of final suffix: koske+ +va+ mietint ¨o+ k a¨si+ +te+ +ll a¨+ + a¨+ (train SMT model on this transformation of training data) (a) Training decoder output: koske+ +va+ mietint o¨+ k a¨si+ +te+ +ll a¨+ + a¨+ decoder output stitched up: koskeva+ mietint o¨+ k ¨asitell ¨a ¨a+ CRF model prediction: x = ‘koskeva+ mietint ¨o+ k ¨asitell ¨a ¨a+’, y = ‘+A +A +n’ koskeva+ +A mietint ¨o+ +A k ¨asitell a¨ ¨a+ +n unstitch morphemes: koske+ +va+ +A mietint ¨o+ +A k ¨asi+ +te+ +ll ¨a+ + ¨a+ +n language model disambiguation: koske+ +va+ +a mietint ¨o+ + a¨ k a¨si+ +te+ +ll a¨+ + a¨+ +n final stitching: koskevaa mietint o¨ ¨a k ¨asitell ¨a ¨an (the output is then compared to the reference translation) (b) Decoding Figure 2: Worked example of all steps in the post-processing morphology prediction model. 3 Experimental Results used the Europarl version 3 corpus (Koehn, 2005) English-Finnish training data set, as well as the standard development and test data sets. Our parallel training data consists of ∼1 million senFor all of the models built in this paper, we tpeanrcaelsle lo tfr a4i0n nwgor ddast or less, sw ohfi ∼le 1t mhei development and test sets were each 2,000 sentences long. In all the experiments conducted in this paper, we used the Moses5 phrase-based translation system (Koehn et al. , 2007) , 2008 version. We trained all of the Moses systems herein using the standard features: language model, reordering model, translation model, and word penalty; in addition to these, the factored experiments called for additional translation and generation features for the added factors as noted above. We used in all experiments the following settings: a hypothesis stack size 100, distortion limit 6, phrase translations limit 20, and maximum phrase length 20. For the language models, we used SRILM 5-gram language models (Stolcke, 2002) for all factors. For our word-based Baseline system, we trained a word-based model using the same Moses system with identical settings. For evaluation against segmented translation systems in segmented forms before word reconstruction, we also segmented the baseline system’s word-based output. All the BLEU scores reported are for lowercase evaluation. We did an initial evaluation of the segmented output translation for each system using the no5http://www.statmt.org/moses/ 36 TabSlBUeuna2gps:meulSipengLmta-e nioatedchMo12dme804-.lB8S714cL±oEr0eUs.6 9 S8up19Nre.358ofe498rUs9ntoihe supervised segmentation baseline model. m-BLEU indicates that the segmented output was evaluated against a segmented version of the reference (this measure does not have the same correlation with human judgement as BLEU) . No Uni indicates the segmented BLEU score without unigrams. tion of m-BLEU score (Luong et al. , 2010) where the BLEU score is computed by comparing the segmented output with a segmented reference translation. Table 2 shows the m-BLEU scores for various systems. We also show the m-BLEU score without unigrams, since over-segmentation could lead to artificially high m-BLEU scores. In fact, if we compare the relative improvement of our m-BLEU scores for the Unsup L-match system we see a relative improvement of 39.75% over the baseline. Luong et. al. (2010) report an m-BLEU score of 55.64% but obtain a relative improvement of 0.6% over their baseline m-BLEU score. We find that when using a good segmentation model, segmentation of the morphologically complex target language improves model performance over an unsegmented baseline (the confidence scores come from bootstrap resampling) . Table 3 shows the evaluation scores for all the baselines and the methods introduced in this paper using standard wordbased lowercase BLEU, WER and PER. We do TSCMaFBU(LubanRolpcesdFotu3lne-ipLr:gMdeLT-tms.al,Stc2ho0r1es:)l 1wB54 Le.r682E90c 27a9Us∗eBL-7 W46E3. U659478R6,1WE-7 TR412E. 847Ra1528nd TER. The ∗ indicates a statistically significant improvement o∗f BndLiEcaUte score over tchalel yB saisgenli nfice mntod imel.The boldface scores are the best performing scores per evaluation measure. better than (Luong et al. , 2010) , the previous best score for this task. We also show a better relative improvement over our baseline when compared to (Luong et al., 2010) : a relative improvement of 4.86% for Unsup L-match compared to our baseline word-based model, compared to their 1.65% improvement over their baseline word-based model. Our best performing method used unsupervised morphology with L-match (see Section 2.2) and the improvement is significant: bootstrap resampling provides a confidence margin of ±0.77 and a t-test (Collins ceot nafli.d , 2005) sahrogwined o significance aw ti-thte p = 0o.0ll0in1s. 3.1 Morphological Fluency Analysis To see how well the models were doing at getting morphology right, we examined several patterns of morphological behavior. While we wish to explore minimally supervised morphological MT models, and use as little language specific information as possible, we do want to use linguistic analysis on the output of our system to see how well the models capture essential morphological information in the target language. So, we ran the word-based baseline system, the segmented model (Unsup L-match) , and the prediction model (CRF-LM) outputs, along with the reference translation through the supervised morphological analyzer Omorfi (Pirinen and Listenmaa, 2007) . Using this analysis, we looked at a variety of linguistic constructions that might reveal patterns in morphological behavior. These were: (a) explicitly marked 37 noun forms, (b) noun-adjective case agreement, (c) subject-verb person/number agreement, (d) transitive object case marking, (e) postpositions, and (f) possession. In each of these categories, we looked for construction matches on a per-sentence level between the models’ output and the reference translation. Table 4 shows the models’ performance on the constructions we examined. In all of the categories, the CRF-LM model achieves the best precision score, as we explain below, while the Unsup L-match model most frequently gets the highest recall score. A general pattern in the most prevalent of these constructions is that the baseline tends to prefer the least marked form for noun cases (corresponding to the nominative) more than the reference or the CRF-LM model. The baseline leaves nouns in the (unmarked) nominative far more than the reference, while the CRF-LM model comes much closer, so it seems to fare better at explicitly marking forms, rather than defaulting to the more frequent unmarked form. Finnish adjectives must be marked with the same case as their head noun, while verbs must agree in person and number with their subject. We saw that in both these categories, the CRFLM model outperforms for precision, while the segmented model gets the best recall. In addition, Finnish generally marks direct objects of verbs with the accusative or the partitive case; we observed more accusative/partitive-marked nouns following verbs in the CRF-LM output than in the baseline, as illustrated by example (1) in Fig. 3. While neither translation picks the same verb as in the reference for the input ‘clarify,’ the CRFLM-output paraphrases it by using a grammatical construction of the transitive verb followed by a noun phrase inflected with the accusative case, correctly capturing the transitive construction. The baseline translation instead follows ‘give’ with a direct object in the nominative case. To help clarify the constructions in question, we have used Google Translate6 to provide back6 http://translate.google. com/ of occurrences per sentence, recall and F-score. also averaged The constructions over the various translations. are listed in descending P, R and F stand for precision, order of their frequency in the texts. The highlighted value in each column is the most accurate with respect to the reference value. translations of our MT output into English; to contextualize these back-translations, we have provided Google’s back-translation of the reference. The use of postpositions shows another difference between the models. Finnish postpositions require the preceding noun to be in the genitive or sometimes partitive case, which occurs correctly more frequently in the CRF-LM than the baseline. In example (2) in Fig. 3, all three translations correspond to the English text, ‘with the basque nationalists. ’ However, the CRF-LM output is more grammatical than the baseline, because not only do the adjective and noun agree for case, but the noun ‘baskien’ to which the postposition ‘kanssa’ belongs is marked with the correct genitive case. However, this well-formedness is not rewarded by BLEU, because ‘baskien’ does not match the reference. In addition, while Finnish may express possession using case marking alone, it has another construction for possession; this can disambiguate an otherwise ambiguous clause. This alternate construction uses a pronoun in the genitive case followed by a possessive-marked noun; we see that the CRF-LM model correctly marks this construction more frequently than the baseline. As example (3) in Fig. 3 shows, while neither model correctly translates ‘matkan’ (‘trip’) , the baseline’s output attributes the inessive ‘yhteydess’ (‘connection’) as belonging to ‘tulokset’ (‘results’) , and misses marking the possession linking it to ‘Commissioner Fischler’. Our manual evaluation shows that the CRF38 LM model is producing output translations that are more morphologically fluent than the wordbased baseline and the segmented translation Unsup L-match system, even though the word choices lead to a lower BLEU score overall when compared to Unsup L-match. 4 Related Work The work on morphology in MT can be grouped into three categories, factored models, segmented translation, and morphology generation. Factored models (Koehn and Hoang, 2007) factor the phrase translation probabilities over additional information annotated to each word, allowing for text to be represented on multiple levels of analysis. We discussed the drawbacks of factored models for our task in Section 2. 1. While (Koehn and Hoang, 2007; Yang and Kirchhoff, 2006; Avramidis and Koehn, 2008) obtain improvements using factored models for translation into English, German, Spanish, and Czech, these models may be less useful for capturing long-distance dependencies in languages with much more complex morphological systems such as Finnish. In our experiments factored models did worse than the baseline. Segmented translation performs morphological analysis on the morphologically complex text for use in the translation model (Brown et al. , 1993; Goldwater and McClosky, 2005; de Gispert and Mari n˜o, 2008) . This method unpacks complex forms into simpler, more frequently occurring components, and may also increase the symmetry of the lexically realized content be(1) Input: ‘the charter we are to approve today both strengthens and gives visible shape to the common fundamental rights and values our community is to be based upon. ’ a. Reference: perusoikeuskirja , jonka t ¨an ¨a ¨an aiomme hyv a¨ksy ¨a , sek ¨a vahvistaa ett ¨a selvent a¨ a¨ (selvent ¨a a¨/VERB/ACT/INF/SG/LAT-clarify) niit a¨ (ne/PRONOUN/PL/PAR-them) yhteisi ¨a perusoikeuksia ja arvoja , joiden on oltava yhteis¨ omme perusta. Back-translation: ‘Charter of Fundamental Rights, which today we are going to accept that clarify and strengthen the common fundamental rights and values, which must be community based. ’ b. Baseline: perusoikeuskirja me hyv ¨aksymme t¨ an ¨a a¨n molemmat vahvistaa ja antaa (antaa/VERB/INF/SG/LATgive) n a¨kyv a¨ (n¨ aky a¨/VERB/ACT/PCP/SG/NOM-visible) muokata yhteist ¨a perusoikeuksia ja arvoja on perustuttava. Back-translation: ‘Charter today, we accept both confirm and modify to make a visible and common values, fundamental rights must be based. ’ c. CRF-LM: perusoikeuskirja on hyv a¨ksytty t ¨an ¨a ¨an , sek ¨a vahvistaa ja antaa (antaa/VERB/ACT/INF/SG/LAT-give) konkreettisen (konkreettinen/ADJECTIVE/SG/GEN,ACC-concrete) muodon (muoto/NOUN/SG/GEN,ACCshape) yhteisi ¨a perusoikeuksia ja perusarvoja , yhteis¨ on on perustuttava. Back-translation: ‘Charter has been approved today, and to strengthen and give concrete shape to the common basic rights and fundamental values, the Community must be based. ’ (2) Input: ‘with the basque nationalists’ a. Reference: baskimaan kansallismielisten kanssa basque-SG/NOM+land-SG/GEN,ACC nationalists-PL/GEN with-POST b. Baseline: baskimaan kansallismieliset kanssa basque-SG/NOM-+land-SG/GEN,ACC kansallismielinen-PL/NOM,ACC-nationalists POST-with c. CRF-LM: kansallismielisten baskien kanssa nationalists-PL/GEN basques-PL/GEN with-POST (3) Input: ‘and in this respect we should value the latest measures from commissioner fischler , the results of his trip to morocco on the 26th of last month and the high level meetings that took place, including the one with the king himself’ a. Reference: ja t ¨ass¨ a mieless ¨a osaamme my¨ os arvostaa komission j¨ asen fischlerin viimeisimpi ¨a toimia , jotka ovat h a¨nen (h¨ anen/GEN-his) marokkoon 26 lokakuuta tekemns (tekem¨ ans ¨a/POSS-his) matkan (matkan/GENtour) ja korkean tason kokousten jopa itsens¨ a kuninkaan kanssa tulosta Back-translation: ‘and in this sense we can also appreciate the Commissioner Fischler’s latest actions, which are his to Morocco 26 October trip to high-level meetings and even the king himself with the result b. Baseline: ja t ¨ass¨ a yhteydess a¨ olisi arvoa viimeisin toimia komission j¨ asen fischler , tulokset monitulkintaisia marokon yhteydess a¨ (yhteydess/INE-connection) , ja viime kuussa pidettiin korkean tason kokouksissa , mukaan luettuna kuninkaan kanssa Back-translation: ‘and in this context would be the value of the last act, Commissioner Fischler, the results of the Moroccan context, ambiguous, and last month held high level meetings, including with the king’ c. CRF-LM: ja t ¨ass¨ a yhteydess a¨ meid ¨an olisi lis ¨aarvoa viimeist ¨a toimenpiteit a¨ kuin komission j¨ asen fischler , ett a¨ h a¨nen (h¨ anen/GEN-his) kokemuksensa (kokemuksensa/POSS-experience) marokolle (marokolle-Moroccan) viime kuun 26 ja korkean tason tapaamiset j¨ arjestettiin, kuninkaan kanssa Back-translation: ‘and in this context, we should value the last measures as the Commissioner Fischler, that his experience in Morocco has on the 26th and high-level meetings took place, including with the king. ’ Figure 3: Morphological fluency analysis (see Section 3. 1) . tween source and target. In a somewhat orthogonal approach to ours, (Ma et al. , 2007) use alignment of a parallel text to pack together adjacent segments in the alignment output, which are then fed back to the word aligner to bootstrap an improved alignment, which is then used in the translation model. We compared our results against (Luong et al. , 2010) in Table 3 since their results are directly comparable to ours. They use a segmented phrase table and language model along with the word-based versions in the decoder and in tuning a Finnish target. Their approach requires segmented phrases 39 to match word boundaries, eliminating morphologically productive phrases. In their work a segmented language model can score a translation, but cannot insert morphology that does not show source-side reflexes. In order to perform a similar experiment that still allowed for morphologically productive phrases, we tried training a segmented translation model, the output of which we stitched up in tuning so as to tune to a word-based reference. The goal of this experiment was to control the segmented model’s tendency to overfit by rewarding it for using correct whole-word forms. However, we found that this approach was less successful than using the segmented reference in tuning, and could not meet the baseline (13.97% BLEU best tuning score, versus 14.93% BLEU for the baseline best tuning score) . Previous work in segmented translation has often used linguistically motivated morphological analysis selectively applied based on a language-specific heuristic. A typical approach is to select a highly inflecting class of words and segment them for particular morphology (de Gispert and Mari n˜o, 2008; Ramanathan et al. , 2009) . Popovi¸ c and Ney (2004) perform segmentation to reduce morphological complexity of the source to translate into an isolating target, reducing the translation error rate for the English target. For Czech-to-English, Goldwater and McClosky (2005) lemmatized the source text and inserted a set of ‘pseudowords’ expected to have lexical reflexes in English. Minkov et. al. (2007) and Toutanova et. al. (2008) use a Maximum Entropy Markov Model for morphology generation. The main drawback to this approach is that it removes morphological information from the translation model (which only uses stems) ; this can be a problem for languages in which morphology ex- presses lexical content. de Gispert (2008) uses a language-specific targeted morphological classifier for Spanish verbs to avoid this issue. Talbot and Osborne (2006) use clustering to group morphological variants of words for word alignments and for smoothing phrase translation tables. Habash (2007) provides various methods to incorporate morphological variants of words in the phrase table in order to help recognize out of vocabulary words in the source language. 5 Conclusion and Future Work We found that using a segmented translation model based on unsupervised morphology induction and a model that combined morpheme segments in the translation model with a postprocessing morphology prediction model gave us better BLEU scores than a word-based baseline. Using our proposed approach we obtain better scores than the state of the art on the EnglishFinnish translation task (Luong et al. , 2010) : from 14.82% BLEU to 15.09%, while using a 40 simpler model. We show that using morphological segmentation in the translation model can improve output translation scores. We also demonstrate that for Finnish (and possibly other agglutinative languages) , phrase-based MT benefits from allowing the translation model access to morphological segmentation yielding productive morphological phrases. Taking advantage of linguistic analysis of the output we show that using a post-processing morphology generation model can improve translation fluency on a sub-word level, in a manner that is not captured by the BLEU word-based evaluation measure. In order to help with replication of the results in this paper, we have run the various morphological analysis steps and created the necessary training, tuning and test data files needed in order to train, tune and test any phrase-based machine translation system with our data. The files can be downloaded from natlang. cs.sfu. ca. In future work we hope to explore the utility of phrases with productive morpheme boundaries and explore why they are not used more pervasively in the decoder. Evaluation measures for morphologically complex languages and tun- ing to those measures are also important future work directions. Also, we would like to explore a non-pipelined approach to morphological preand post-processing so that a globally trained model could be used to remove the target side morphemes that would improve the translation model and then predict those morphemes in the target language. Acknowledgements This research was partially supported by NSERC, Canada (RGPIN: 264905) and a Google Faculty Award. We would like to thank Christian Monson, Franz Och, Fred Popowich, Howard Johnson, Majid Razmara, Baskaran Sankaran and the anonymous reviewers for their valuable comments on this work. We would particularly like to thank the developers of the open-source Moses machine translation toolkit and the Omorfi morphological analyzer for Finnish which we used for our experiments. References Eleftherios Avramidis and Philipp Koehn. 2008. Enriching morphologically poor languages for statistical machine translation. In Proceedings of the 46th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, page 763?770, Columbus, Ohio, USA. Association for Computational Linguistics. Peter F. Brown, Stephen A. Della Pietra, Vincent J. Della Pietra, and R. L. Mercer. 1993. The mathematics of statistical machine translation: Parameter estimation. Computational Linguistics, 19(2) :263–31 1. Pi-Chuan Chang, Michel Galley, and Christopher D. Manning. 2008. Optimizing Chinese word segmentation for machine translation performance. In Proceedings of the Third Workshop on Statistical Machine Translation, pages 224–232, Columbus, Ohio, June. Association for Computational Linguistics. Michael Collins, Philipp Koehn, and Ivona Kucerova. 2005. Clause restructuring for statistical machine translation. In Proceedings of 43rd Annual Meeting of the Association for Computational Linguistics (A CL05). Association for Computational Linguistics. Mathias Creutz and Krista Lagus. 2005. Inducing the morphological lexicon of a natural language from unannotated text. In Proceedings of the International and Interdisciplinary Conference on Adaptive Knowledge Representation and Reason- ing (AKRR ’05), pages 106–113, Espoo, Finland. Mathias Creutz and Krista Lagus. 2006. Morfessor in the morpho challenge. In Proceedings of the PASCAL Challenge Workshop on Unsupervised Segmentation of Words into Morphemes. Adri ´a de Gispert and Jos e´ Mari n˜o. 2008. On the impact of morphology in English to Spanish statistical MT. Speech Communication, 50(11-12) . Sharon Goldwater and David McClosky. 2005. Improving statistical MT through morphological analysis. In Proceedings of the Human Language Technology Conference and Conference on Empirical Methods in Natural Language Processing, pages 676–683, Vancouver, B.C. , Canada. Association for Computational Linguistics. Philipp Koehn and Hieu Hoang. 2007. Factored translation models. In Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 868–876, Prague, Czech Republic. Association for Computational Linguistics. 41 Philipp Koehn, Hieu Hoang, Alexandra Birch, Chris Callison-Burch, Marcello Federico, Nicola Bertoldi, Brooke Cowan, Wade Shen, Christine Moran, Richard Zens, Chris Dyer, Ondrej Bojar, Alexandra Constantin, and Evan Herbst. 2007. Moses: Open source toolkit for statistical machine translation. In A CL ‘07: Proceedings of the 45th Annual Meeting of the A CL on Interactive Poster and Demonstration Sessions, pages 177–108, Prague, Czech Republic. Association for Computational Linguistics. Philipp Koehn. 2005. Europarl: A parallel corpus for statistical machine translation. In Proceedings of Machine Translation Summit X, pages 79–86, Phuket, Thailand. Association for Computational Linguistics. John Lafferty, Andrew McCallum, and Fernando Pereira. 2001. Conditional random fields: Probabilistic models for segmenting and labeling sequence data. In Proceedings of the 18th International Conference on Machine Learning, pages 282–289, San Francisco, California, USA. Association for Computing Machinery. Minh-Thang Luong, Preslav Nakov, and Min-Yen Kan. 2010. A hybrid morpheme-word representation for machine translation of morphologically rich languages. In Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 148–157, Cambridge, Massachusetts. Association for Computational Linguistics. Yanjun Ma, Nicolas Stroppa, and Andy Way. 2007. Bootstrapping word alignment via word packing. In Proceedings of the 45th Annual Meeting of the Association of Computational Linguistics, pages 304–311, Prague, Czech Republic. Association for Computational Linguistics. Einat Minkov, Kristina Toutanova, and Hisami Suzuki. 2007. Generating complex morphology for machine translation. In In Proceedings of the 45th Annual Meeting of the Association for Computational Linguistics (A CL07), pages 128–135, Prague, Czech Republic. Association for Computational Linguistics. Christian Monson. 2008. Paramor and morpho challenge 2008. In Lecture Notes in Computer Science: Workshop of the Cross-Language Evaluation Forum (CLEF 2008), Revised Selected Papers. Habash Nizar. 2007. Four techniques for online handling of out-of-vocabulary words in arabic-english statistical machine translation. In Proceedings of the 46th Annual Meeting of the Association of Computational Linguistics, Columbus, Ohio. Association for Computational Linguistics. Kishore Papineni, Salim Roukos, Todd Ward, and Zhu. 2002. BLEU: A method for automatic evaluation of machine translation. In Proceedings of 40th Annual Meeting of the Association for Computational Linguistics A CL, pages 311–318, Philadelphia, Pennsylvania, USA. Association for Computational Linguistics. Tommi Pirinen and Inari Listenmaa. 2007. Omorfi morphological analzer. http://gna.org/projects/omorfi. Maja Popovi¸ c and Hermann Ney. 2004. Towards the use of word stems and suffixes for statistiWei jing cal machine translation. In Proceedings of the 4th International Conference on Language Resources and Evaluation (LREC), pages 1585–1588, Lisbon, Portugal. European Language Resources Association (ELRA) . Ananthakrishnan Ramanathan, Hansraj Choudhary, Avishek Ghosh, and Pushpak Bhattacharyya. 2009. Case markers and morphology: Addressing the crux of the fluency problem in EnglishHindi SMT. In Proceedings of the Joint Conference of the 4 7th Annual Meeting of the Association for Computational Linguistics and the 4th International Joint Conference on Natural Language Processing of the Asian Federation of Natural Language Processing, pages 800–808, Suntec, Singapore. Association for Computational Linguistics. Andreas Stolcke. 2002. Srilm – an extensible language modeling toolkit. 7th International Conference on Spoken Language Processing, 3:901–904. David Talbot and Miles Osborne. 2006. Modelling lexical redundancy for machine translation. In Proceedings of the 21st International Conference on Computational Linguistics and 44th Annual Meeting of the Association for Computational Linguistics, pages 969–976, Sydney, Australia, July. Association for Computational Linguistics. Kristina Toutanova, Hisami Suzuki, and Achim Ruopp. 2008. Applying morphology generation models to machine translation. In Proceedings of the 46th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, pages 514–522, Columbus, Ohio, USA. Association for Computational Linguistics. Mei Yang and Katrin Kirchhoff. 2006. Phrase-based backoff models for machine translation of highly inflected languages. In Proceedings of the European Chapter of the Association for Computational Linguistics, pages 41–48, Trento, Italy. Association for Computational Linguistics. 42

5 0.83579224 211 acl-2011-Liars and Saviors in a Sentiment Annotated Corpus of Comments to Political Debates

Author: Paula Carvalho ; Luis Sarmento ; Jorge Teixeira ; Mario J. Silva

Abstract: We investigate the expression of opinions about human entities in user-generated content (UGC). A set of 2,800 online news comments (8,000 sentences) was manually annotated, following a rich annotation scheme designed for this purpose. We conclude that the challenge in performing opinion mining in such type of content is correctly identifying the positive opinions, because (i) they are much less frequent than negative opinions and (ii) they are particularly exposed to verbal irony. We also show that the recognition of human targets poses additional challenges on mining opinions from UGC, since they are frequently mentioned by pronouns, definite descriptions and nicknames. 1

6 0.82456219 40 acl-2011-An Error Analysis of Relation Extraction in Social Media Documents

7 0.8238948 216 acl-2011-MEANT: An inexpensive, high-accuracy, semi-automatic metric for evaluating translation utility based on semantic roles

8 0.82053274 301 acl-2011-The impact of language models and loss functions on repair disfluency detection

9 0.76622313 274 acl-2011-Semi-Supervised Frame-Semantic Parsing for Unknown Predicates

10 0.75979447 286 acl-2011-Social Network Extraction from Texts: A Thesis Proposal

11 0.75785542 3 acl-2011-A Bayesian Model for Unsupervised Semantic Parsing

12 0.75431776 84 acl-2011-Contrasting Opposing Views of News Articles on Contentious Issues

13 0.75092608 310 acl-2011-Translating from Morphologically Complex Languages: A Paraphrase-Based Approach

14 0.74991131 62 acl-2011-Blast: A Tool for Error Analysis of Machine Translation Output

15 0.7495631 149 acl-2011-Hierarchical Reinforcement Learning and Hidden Markov Models for Task-Oriented Natural Language Generation

16 0.74903333 37 acl-2011-An Empirical Evaluation of Data-Driven Paraphrase Generation Techniques

17 0.74839175 21 acl-2011-A Pilot Study of Opinion Summarization in Conversations

18 0.74758023 32 acl-2011-Algorithm Selection and Model Adaptation for ESL Correction Tasks

19 0.74708551 137 acl-2011-Fine-Grained Class Label Markup of Search Queries

20 0.74615896 128 acl-2011-Exploring Entity Relations for Named Entity Disambiguation