If user interfaces should be made human-like vs. tool-like has been debated in the HCI field, and this debate affects the development of multimodal interfaces. However, little empirical study has been done to support either view so far. Even if there is evidence that humans interpret media as other humans, this does not mean that humans experience the interfaces as human-like. We studied how people experience a multimodal timetable system with varying degree of human-like spoken feedback in a Wizard-of-Oz study. The results showed that users' views and preferences lean significantly towards anthropomorphism after actually experiencing the multimodal timetable system. The more human-like the spoken feedback is the more participants preferred the system to be human-like. The results also showed that the users experience matched their preferences. This shows that in order to appreciate a human-like interface, the users have to experience it.

Using natural language in addition to graphical user interfaces is often used as an argument for a better interaction. However, just adding spoken language might not lead to a better interaction. In this article we will look deeper into how the spoken language should be used in a co-operative multimodal interface. Based on empirical investigations, we have noticed that for multimodal information systems efficiency is especially important. Our results indicate that efficiency can be divided into functional and linguistic efficiency. Functional efficiency has a tight relation to solving the task fast. Linguistic efficiency concerns how to make the contributions meaningful and appropriate in the context. For linguistic efficiency user's perception of first-personness [1] is important, as well as giving users support for understanding the interface, and to adapt the responses to the user. In this article focus is on linguistic efficiency for a multimodal timetable information system.

This volume is unique in its breadth of coverage on key topics in the field from a variety of leading researchers. In one volume, readers gain exposure to several perspectives in the areas of corpus annotation and analysis, dialogue system construction; as well as theoretical perspectives on communicative intention, context-based generation, and modelling of discourse structure. In this book you will find high quality articles representing current and new directions in discourse and dialogue with an emphasis on Dialogue Systems; Corpora and Corpus Tools; and Semantic and Pragmatic Modelling of Discourse and Dialogue. The majority of the articles included come from the most outstanding papers presented at the 2^nd SIGdial workshop on Discourse and Dialogue held in conjunction with Eurospeech 2001. The contents are supplemented with four invited papers from internationally recognized researchers in discourse and dialogue.

Speech communication involves normally not only speech but also face and head movements. In the present investigation the visual correlates to focal accent in read speech and to confirmation in Swedish are studied and a computational model for the movements is hypothesized. Focal accent is signalling "new" information in speech and is signalled by means of the fundamental frequency manifestation and by prolonged segment durations. The head movements are recorded by the Qualisys MacReflex motion tracking system simultaneously with the speech signal. The results show that head movements that eo-occur with the signalling of focal accent in the speech signal will have the extreme values at the primary stressed syllable of the word carrying focal accent independent of the word accent type in Swedish. To be noted is that focal accent in Swedish will have the fundamental frequency manifestation in words carrying the word accent II at the secondary stressed vowel. The nodding that is signalling confirmation is signalled by means of a damped oscillation of the head. The head movements in both cases may be simulated by a second order linear system and the different patterns are two of the three possible solutions to the equations

Speech communication normally involves not only speech but also face and head movements. In the present investigation, the correlation between head movement and focus assignment is studied, both in the laboratory and in spontaneous speech, with the aim of finding out what these head movements look like in detail. Specifically addressed questions are whether the head movements are an obligatory signal of focus assignment, and in that case how often a head movement will accompany the prosodic information. Also studied are where in the focused word the head movement has its extreme value, the relationship of that value to the extreme value of the fundamental frequency, and whether it is possible to simulate the head movements that accompany focal accent with a secondary order linear system.

In this study, the head movements are recorded by the Qualisys MacReflex motion tracking system simultaneously with the speech signal. The results show that, for the subjects studied, the head movements that coincide with the signalling of focal accent in the speech signal, in most cases, have their extreme values at the primary stressed syllable of the word carrying focal accent, independent of the word accent type in Swedish. It should be noted that focal accent in Swedish has the fundamental frequency manifestation in words carrying the word accent II on the secondary stressed vowel.

The time required for the head movement to reach the extreme value is longer than the corresponding time for the fundamental frequency rise probably due to the mass of the head in comparison to the structure involved for the fundamental frequency manipulation. The head movements are simulated with a high accuracy by a second order linear system.

This paper reports results from a comparative analysis of purring in four domesticcats. An acoustic analysis describes sound pressure level, duration, number ofcycles and fundamental frequency for egressive and ingressive phases. Significantindividual differences are found between the four cats in several respects.

In this paper we present results from using Random indexing for Latent Semantic Analysis to handle Singular Value Decomposition tractability issues. In the paper we compare Latent Semantic Analysis, Random Indexing and Latent Semantic Analysis on Random Indexing reduced matrices. Our results show that Latent Semantic Analysis on Random Indexing reduced matrices provide better results on Precision and Recall than Random Indexing only. Furthermore, computation time for Singular Value Decomposition on a Random indexing reduced matrix is almost halved compared to Latent Semantic Analysis.

The paper discusses the motivation for and outcome of the addition of socially oriented so called “off-task” conversational abilities to an existing “teachable agent” (TA) in an educational mathematics game. The purpose of the extension is to affect constructs known to promote learning, such as self-efficacy and engagement, as well as enhancing students’ experiences of the game. A comparison of students that played the educational game with the off-task interaction included to those who played without it, indicate that the former had a more positive experience of the game, and that they also learnt more in the sense of teaching their TA better. The potential for pedagogical interventions in this and similar systems is discussed as well as differences found between high- and low-achievers.

We are developing a virtual learning environment, which includes a math game with a teachable agent, the embodiment of the agent, and a social conversation with the agent. In this work we are using various methods for design and system development, with a focus on iterative methods, and high involvement of the user in the process. In this paper we discuss the motivation for and the applications of these methods in the development of the social conversation with the agent.

In this paper we discuss how social conversation with an agent in an educational math game can be used in order to gain pedagogical benefits, for example to increase positive attitudes towards learning math and math self-efficacy. We present the iterative development of the conversational module, architectural considerations, and the type of dialogue phenomenon that support the pedagogical interventions

We describe our work-in-progress of developing an educational game in mathematics for 12-14 year olds, by adding social and conversational abilities to an existing “teachable agent” (TA) in the game. The purpose of this extension is to affect cognitive, emotional and social constructs known to promote learning, such as self-efficacy and engagement, as well as enhancing students’ experiences of interacting with the agent over an extended period of time. Drawing from the EnALI framework, which states practical design guidelines, we discuss specific design challenges and exemplify research considerations as to developing the agent’s visual representation and conversational module. We present some initial findings from using prototype agents with students from the target group. Promising developments seem to reside in pronouncing the agent’s personality traits and expanding its knowledge database, particularly its range of conversational topics. Finally we propose some future studies and research directions.

This thesis explores the possibility of assessing the degree of readability of writtenSwedish using machine learning. An application using four levels of linguistic analysishas been implemented and tested with four different established algorithmsfor machine learning. The new approach has then been compared to establishedreadability metrics for Swedish. The results indicate that the new method workssignificantly better for readability classification of both sentences and documents.The system has also been tested with so called soft classification which returns aprobability for the degree of readability of a given text. This probability can thenbe used to rank texts according to probable degree of readability.

In this article, compound processing for translation into German in a factored statistical MT system is investigated. Compound sare handled by splitting them prior to training, and merging the parts after translation. I have explored eight merging strategies using different combinations of external knowledge sources, such as word lists, and internal sources that are carried through the translation process, such as symbols or parts-of-speech. I show that for merging to be successful, some internal knowledge source is needed. I also show that an extra sequence model for part-ofspeech is useful in order to improve the order of compound parts in the output. The best merging results are achieved by a matching scheme for part-of-speech tags.

We present BLAST, an open source tool for error analysis of machine translation (MT) output. We believe that error analysis, i.e., to identify and classify MT errors, should be an integral part of MT development, since it gives a qualitative view, which is not obtained by standard evaluation methods. BLAST can aid MT researchers and users in this process, by providing an easy-to-use graphical user interface. It is designed to be flexible, and can be used with any MT system, language pair, and error typology. The annotation task can be aided by highlighting similarities with a reference translation.

Clustered word classes have been used in connection with statistical machine translation, for instance for improving word alignments. In this work we investigate if clustered word classes can be used in a preordering strategy, where the source language is reordered prior to training and translation. Part-of-speech tagging has previously been successfully used for learning reordering rules that can be applied before training and translation. We show that we can use word clusters for learning rules, and significantly improve on a baseline with only slightly worse performance than for standard POS-tags on an English–German translation task. We also show the usefulness of the approach for the less-resourced language Haitian Creole, for translation into English, where the suggested approach is significantly better than the baseline.

Translation into compounding languages like German and Swedish is a challenge for statistical machine translation. I present a novel algorithm for merging compound parts, based on part-of-speech matching with an extended tag set. It improves the quality of merged compounds compared to previously suggested methods, both measured automatically and shown in an error analysis. Translation is also improved compared to systems without compound processing for Swedish, Danish,  and German.

In this thesis I explore how compound processing can be used to improve phrase-based statistical machine translation (PBSMT) between English and German/Swedish. Both German and Swedish generally use closed compounds, which are written as one word without spaces or other indicators of word boundaries. Compounding is both common and productive, which makes it problematic for PBSMT, mainly due to sparse data problems.

The adopted strategy for compound processing is to split compounds into their component parts before training and translation. For translation into Swedish and German the parts are merged after translation. I investigate the effect of different splitting algorithms for translation between English and German, and of different merging algorithms for German. I also apply these methods to a different language pair, English--Swedish. Overall the studies show that compound processing is useful, especially for translation from English into German or Swedish. But there are improvements for translation into English as well, such as a reduction of unknown words.

I show that for translation between English and German different splitting algorithms work best for different translation directions. I also design and evaluate a novel merging algorithm based on part-of-speech matching, which outperforms previous methods for compound merging, showing the need for information that is carried through the translation process, rather than only external knowledge sources such as word lists. Most of the methods for compound processing were originally developed for German. I show that these methods can be applied to Swedish as well, with similar results.

There are two ways to express definiteness in Danish, which makes it problematic for statistical machine translation (SMT) from English, since the wrong realisation can be chosen. We present a part-of-speech-based method for identifying and transforming English definite NPs that would likely be expressed in a different way in Danish. The transformed English is used for training a phrase-based SMT system.This technique gives significant improvements of translation quality, of up to 22.1% relative on Bleu, compared to a baseline trained on original English, in two different domains.

The Scandinavian languages have an unusual structure of definite noun phrases (NPs), with a noun suffix as one possibility of expressing definiteness, which is problematic for statistical machine translation from languages with different NP structures. We show that translation can be improved by simple source side transformations of definite NPs, for translation from English and Italian, into Danish, Swedish, and Norwegian, with small adjustments of the preprocessing strategy, depending on the language pair. We also explored target side transformations, with mixed results.

We have conducted initial experiments with statistical machine translation between English and Swedish based on the Moses toolkit and the Europarl corpus. The main aim was to decrease processing times without harming translation quality by changing the settings in Moses and for the parameter tuning. The experiments show that translation and tuning times can be cut around 10 times without harming translation quality, and in some cases the quality is even increased.

An empirical method for splitting German compounds is explored by varying it in a number of ways to investigate the consequences for factored statistical machine translation between English and German in both directions. Compound splitting is incorporated into translation in a preprocessing step, performed on training data and on German translation input. For translation into German, compounds are merged based on part-of-speech in a postprocessing step. Compound parts are marked, to separate them from ordinary words. Translation quality is improved in both translation directions and the number of untranslated words in the English output is reduced. Different versions of the splitting algorithm performs best in the two different translation directions.

Word alignment is necessary for statistical machine translation (SMT), and reordering as a preprocessing step has been shown to improve SMT for many language pairs. In this initial study we investigate if both word alignment and reordering can be improved by iterating these two steps, since they both depend on each other. Overall no consistent improvements were seen on the translation task, but the reordering rules contain different information in the different iterations, leading us to believe that the iterative strategy can be useful.

In this thesis proposal I present my thesis work, about pre- and postprocessing for statistical machine translation, mainly into Germanic languages. I focus my work on four areas: compounding, definite noun phrases, reordering, and error correction. Initial results are positive within all four areas, and there are promising possibilities for extending these approaches. In addition I also focus on methods for performing thorough error analysis of machine translation output, which can both motivate and evaluate the studies performed.

We report results on translation of SMS messages from Haitian Creole to English. We show improvements by applying spell checking techniques to unknown words and creating a lattice with the best known spelling equivalents. We also used a small cleaned corpus to train a cleaning model that we applied to the noisy corpora.

In this thesis I aim to improve phrase-based statistical machine translation (PBSMT) in a number of ways by the use of text harmonization strategies. PBSMT systems are built by training statistical models on large corpora of human translations. This architecture generally performs well for languages with similar structure. If the languages are different for example with respect to word order or morphological complexity, however, the standard methods do not tend to work well. I address this problem through text harmonization, by making texts more similar before training and applying a PBSMT system.

I investigate how text harmonization can be used to improve PBSMT with a focus on four areas: compounding, definiteness, word order, and unknown words. For the first three areas, the focus is on linguistic differences between languages, which I address by applying transformation rules, using either rule-based or machine learning-based techniques, to the source or target data. For the last area, unknown words, I harmonize the translation input to the training data by replacing unknown words with known alternatives.

I show that translation into languages with closed compounds can be improved by splitting and merging compounds. I develop new merging algorithms that outperform previously suggested algorithms and show how part-of-speech tags can be used to improve the order of compound parts. Scandinavian definite noun phrases are identified as a problem forPBSMT in translation into Scandinavian languages and I propose a preprocessing approach that addresses this problem and gives large improvements over a baseline. Several previous proposals for how to handle differences in reordering exist; I propose two types of extensions, iterating reordering and word alignment and using automatically induced word classes, which allow these methods to be used for less-resourced languages. Finally I identify several ways of replacing unknown words in the translation input, most notably a spell checking-inspired algorithm, which can be trained using character-based PBSMT techniques.

Overall I present several approaches for extending PBSMT by the use of pre- and postprocessing techniques for text harmonization, and show experimentally that these methods work. Text harmonization methods are an efficient way to improve statistical machine translation within the phrase-based approach, without resorting to more complex models.

Error analysis is a means to assess machine translation output in qualitative terms, which can be used as a basis for the generation of error profiles for different systems. As for other subjective approaches to evaluation it runs the risk of low inter-annotator agreement, but very often in papers applying error analysis to MT, this aspect is not even discussed. In this paper, we report results from a comparative evaluation of two systems where agreement initially was low, and discuss the different ways we used to improve it. We compared the effects of using more or less fine-grained taxonomies, and the possibility to restrict analysis to short sentences only. We report results on inter-annotator agreement before and after measures were taken, on error categories that are most likely to be confused, and on the possibility to establish error profiles also in the absence of a high inter-annotator agreement.

One problem in statistical machine translation (SMT) is that the output often is ungrammatical. To address this issue, we have investigated the use of a grammar checker for two purposes in connection with SMT: as an evaluation tool and as a postprocessing tool. As an evaluation tool the grammar checker gives a complementary picture to standard metrics such as Bleu, which do not account for grammaticality. We use the grammar checker as a postprocessing tool by applying the error correction suggestions it gives. There are only small overall improvements of the postprocessing on automatic metrics, but the sentences that are affected by the changes are improved, as shown both by automatic metrics and by a human error analysis. These results indicate that grammar checker techniques are a useful complement to SMT.

In many languages the use of compound words is very productive. A common practice to reduce sparsity consists in splitting compounds in the training data. When this is done, the system incurs the risk of translating components in non-consecutive positions, or in the wrong order. Furthermore, a post-processing step of compound merging is required to reconstruct compound words in the output. We present a method for increasing the chances that components that should be merged are translated into contiguous positions and in the right order. We also propose new heuristic methods for merging components that outperform all known methods, and a learning-based method that has similar accuracy as the heuristic method, is better at producing novel compounds, and can operate with no background linguistic resources.

In this article we investigate statistical machine translation (SMT) into Germanic languages, with a focus on compound processing. Our main goal is to enable the generation of novel compounds that have not been seen in the training data. We adopt a split-merge strategy, where compounds are split before training the SMT system, and merged after the translation step. This approach reduces sparsity in the training data, but runs the risk of placing translations of compound parts in non-consecutive positions. It also requires a postprocessing step of compound merging, where compounds are reconstructed in the translation output. We present a method for increasing the chances that components that should be merged are translated into contiguous positions and in the right order and show that it can lead to improvements both by direct inspection and in terms of standard translation evaluation metrics. We also propose several new methods for compound merging, based on heuristics and machine learning, which outperform previously suggested algorithms. These methods can produce novel compounds and a translation with at least the same overall quality as the baseline. For all subtasks we show that it is useful to include part-of-speech based information in the translation process, in order to handle compounds.

We investigated the effects of processing Swedish compounds for phrase-based SMT between Swedish and English. Compounds were split in a pre-processing step using an unsupervised empirical method. After translation into Swedish, compounds were merged, using a novel merging algorithm. We investigated two ways of handling compound parts, by marking them as compound parts or by normalizing them to a canonical form. We found that compound splitting did improve translation into Swedish, according to automatic metrics. For translation into English the results were not consistent across automatic metrics. However, error analysis of compound translation showed a small improvement in the systems that used splitting. The number of untranslated words in the English output was reduced by 50%.

We describe the LIU systems for German-English and English-German translation submitted to the Shared Task of the Third Workshop of Statistical Machine Translation. The main features of the systems, as compared with the baseline, is the use of morphological pre- and post-processing, and a sequence model for German using morphologically richparts-of-speech. It is shown that these additions lead to improved translations.

In this paper we report on experiments with three preprocessing strategies for improving translation output in a statistical MT system. In training, two reordering strategies were studied: (i) reorder on thebasis of the alignments from Giza++, and (ii) reorder by moving all verbs to the end of segments. In translation, out-of-vocabulary words were preprocessed in a knowledge-lite fashion to identify a likely equivalent. All three strategies were implemented for our English-German systems submitted to the WMT10 shared task. Combining them lead to improvements in both language directions.

Question answering systems can be seen as the next step in information retrieval, allowing users to pose questions in natural language and receive succinct answers. In order for a question answering system as a whole to be successful, research has shown that the correct classification of questions with regards to the expected answer type is imperative. Question classification has two components: a taxonomy of answer types, and a machinery for making the classifications.

This thesis focuses on five different machine learning algorithms for the question classification task. The algorithms are k nearest neighbours, naïve bayes, decision tree learning, sparse network of winnows, and support vector machines. These algorithms have been applied to two different corpora, one of which has been used extensively in previous work and has been constructed for a specific agenda. The other corpus is drawn from a set of users' questions posed to a running online system. The results showed that the performance of the algorithms on the different corpora differs both in absolute terms, as well as with regards to the relative ranking of them. On the novel corpus, naïve bayes, decision tree learning, and support vector machines perform on par with each other, while on the biased corpus there is a clear difference between them, with support vector machines being the best and naïve bayes being the worst.

The thesis also presents an analysis of questions that are problematic for all learning algorithms. The errors can roughly be divided as due to categories with few members, variations in question formulation, the actual usage of the taxonomy, keyword errors, and spelling errors. A large portion of the errors were also hard to explain.

In this paper we present a novel technique to capture Web users’ behaviour based on their interest-oriented actions. In our approach we utilise the vector space model Random Indexing to identify the latent factors or hidden relationships among Web users’ navigational behaviour. Random Indexing is an incremental vector space technique that allows for continuous Web usage mining. User requests are modelled by Random Indexing for individual users’ navigational pattern clustering and common user profile creation. Clustering Web users’ access patterns may capture common user interests and, in turn, build user profiles for advanced Web applications, such as Web caching and prefetching. We present results from the Web user clustering approach through experiments on a real Web log file with promising results. We also apply our data to a prefetching task and compare that with previous approaches. The results show that Random Indexing provides more accurate prefetchings.

This thesis examines recommendation dialogue, in the context of dialogue strategy design for conversational recommender systems. The purpose of a recommender system is to produce personalized recommendations of potentially useful items from a large space of possible options. In a conversational recommender system, this task is approached by utilizing natural language recommendation dialogue for detecting user preferences, as well as for providing recommendations. The fundamental idea of a conversational recommender system is that it relies on dialogue sessions to detect, continuously update, and utilize the user's preferences in order to predict potential interest in domain items modeled in a system. Designing the dialogue strategy management is thus one of the most important tasks for such systems.

Based on empirical studies as well as design and implementation of conversational recommender systems, a behavior-based dialogue model called bcorn is presented. bcorn is based on three constructs, which are presented in the thesis. It utilizes a user preference modeling framework (preflets) that supports and utilizes natural language dialogue, and allows for descriptive, comparative, and superlative preference statements, in various situations. Another component of bcorn is its message-passing formalism, pcql, which is a notation used when describing preferential and factual statements and requests. bcorn is designed to be a generic recommendation dialogue strategy with conventional, information-providing, and recommendation capabilities, that each describes a natural chunk of a recommender agent's dialogue strategy, modeled in dialogue behavior diagrams that are run in parallel to give rise to coherent, flexible, and effective dialogue in conversational recommender systems.

Three empirical studies have been carried out in order to explore the problem space of recommendation dialogue, and to verify the solutions put forward in this work. Study I is a corpus study in the domain of movie recommendations. The result of the study is a characterization of recommendation dialogue, and forms a base for a first prototype implementation of a human-computer recommendation dialogue control strategy. Study II is an end-user evaluation of the acorn system that implements the dialogue control strategy and results in a verification of the effectiveness and usability of the dialogue strategy. There are also implications that influence the refinement of the model that are used in the bcorn dialogue strategy model. Study III is an overhearer evaluation of a functional conversational recommender system called CoreSong, which implements the bcorn model. The result of the study is indicative of the soundness of the behavior-based approach to conversational recommender system design, as well as the informativeness, naturalness, and coherence of the individual bcorn dialogue behaviors.

My research interest lies in investigating user-adaptive interaction in a conversational setting for recommender systems, with particular focus on modularized user model components and the use of a dialogue partner (DP) in such systems.