Messages exchanged between participants in online discussion forums often contain personal names and other details that need to be redacted before the data is used for research purposes in learning analytics. However, removing the names entirely makes it harder to track the exchange of ideas between individuals within a message thread and across threads, and thereby reduces the value of this type of conversational data. In contrast, the consistent use of pseudonyms allows contributions from individuals to be tracked across messages, while also hiding the real identities of the contributors. Several factors can make it difficult to identify all instances of personal names that refer to the same individual, including spelling errors and the use of shortened forms. We developed a semi-automated approach for replacing personal names with consistent pseudonyms. We evaluated our approach on a data set of over 1,700 messages exchanged during a distance-learning course, and compared it to a general-purpose pseudonymisation tool that used deep neural networks to identify names to be redacted. We found that our tailored approach out-performed the general-purpose tool in both precision and recall, correctly identifying all but 31 substitutions out of 2,888.

Writing from multiple sources is a commonly administered learning task across educational levels and disciplines. In this task, learners are instructed to comprehend information from source documents and integrate it into a coherent written composition to fulfil the assignment requirements. Even though educationally potent, multi-source writing tasks are considered challenging to many learners, in particular because many learners underuse monitoring and control, critical metacognitive processes for productive engagement in multi-source writing. To understand these processes, we conducted a laboratory study involving 44 university students. They engaged in multi-source writing task hosted in digital learning environment. Adding to previous research, we unobtrusively measured metacognitive processes using learners’ trace data collected via multiple data channels and in both writing and reading space of the multi-source writing task. We further investigated how these processes affect the quality of a written product, i.e., essay score. In the analysis, we utilised both automatically and human-generated essay score. The rating performance of the essay scoring algorithm was comparable to that of human raters. Our results largely support the theoretical assumptions that engagement in metacognitive monitoring and control benefits the quality of written product. Moreover, our results can inform the development of analytics-based tools that support student writing by making use of trace data and automated essay scoring.

Self-regulated learning (SRL) skills are essential for successful learning in a technology-enhanced learning environment. Learning Analytics techniques have shown a great potential in identifying and exploring SRL strategies from trace data in various learning environments. However, these strategies have been mainly identified through analysis of sequences of learning actions, and thus interpretation of the strategies is heavily task and context dependent. Further, little research has been done on the association of SRL strategies with different influencing factors or conditions. To address these gaps, we propose an analytic method for detecting SRL strategies from theoretically supported SRL processes and applied the method to a dataset collected from a multi-source writing task. The detected SRL strategies were explored in terms of their association with the learning outcome, internal conditions (prior-knowledge, metacognitive knowledge and motivation) and external conditions (scaffolding). The study results showed our analytic method successfully identified three theoretically meaningful SRL strategies. The study results revealed small effect size in the association between the internal conditions and the identified SRL strategies, but revealed a moderate effect size in the association between external conditions and the SRL strategy use.

Asynchronous online discussions within a community of learners can improve learning outcomes through social knowledge construction, but the depth and quality of student contributions often varies widely. Approaches to assessing critical discourse typically use content analysis to identify indicators that correspond to framework constructs, that in turn serve as measures of depth and quality. Often only a single construct is addressed for performing content analysis in the literature, although recent work has used both social presence and cognitive presence constructs from the Community of Inquiry (CoI) framework. Nevertheless, there is no effective, commonly used, analytic approach to combining insights from multiple perspectives about quality and depth of online discussions. This paper addresses the gap by proposing the combined use of cognitive engagement (the ICAP framework) and cognitive presence (CoI); and by proposing a network analytic approach that quantifies the associations between the two frameworks and measures the moderation effects of two instructional interventions on those associations. The present study found that these associations were moderated by one intervention but not the other; and that messages labelled with the most common phase of cognitive presence could be usefully assigned to smaller meaningful subgroups by also considering the mode of cognitive engagement.

Researchers have been struggling with the measurement of Self-Regulated Learning (SRL) for decades. Instrumentation tools have been proposed to help capture SRL processes that are difficult to capture. The aim of the present study was to improve measurement of SRL by embedding instrumentation tools in a learning environment and validating the measurement of SRL with these instrumentation tools using think aloud. Synchronizing log data and concurrent think aloud data helped identify which SRL processes were captured by particular instrumentation tools. One tool was associated with a single SRL process: the timer co-occurred with monitoring. Other tools co-occurred with a number of SRL processes, i.e., the highlighter and note taker captured superficial writing down, organizing, and monitoring, whereas the search and planner tools revealed planning and monitoring. When specific learner actions with the tool were analyzed, a clearer picture emerged of the relation between the highlighter and note taker and SRL processes. By aligning log data with think aloud data, we showed that instrumentation tool use indeed reflects SRL processes. The main contribution is that this paper is the first to show that SRL processes that are difficult to measure by trace data can indeed be captured by instrumentation tools such as high cognition and metacognition. Future challenges are to collect and process log data real time with learning analytic techniques to measure ongoing SRL processes and support learners during learning with personalized SRL scaffolds.

Recognizing emotional reactions of movie audiences to affective movie content is a challenging task in affective computing. Previous research on induced emotion recognition has mainly focused on using audio-visual movie content. Nevertheless, the relationship between the perceptions of the affective movie content (perceived emotions) and the emotions evoked in the audiences (induced emotions) is unexplored. In this work, we studied the relationship between perceived and induced emotions of movie audiences. Moreover, we investigated multimodal modelling approaches to predict movie induced emotions from movie content based features, as well as physiological and behavioral reactions of movie audiences. To carry out analysis of induced and perceived emotions, we first extended an existing database for movie affect analysis by annotating perceived emotions in a crowd-sourced manner. We find that perceived and induced emotions are not always consistent with each other. In addition, we show that perceived emotions, movie dialogues, and aesthetic highlights are discriminative for movie induced emotion recognition besides spectators' physiological and behavioral reactions. Also, our experiments revealed that induced emotion recognition could benefit from including temporal information and performing multimodal fusion. Moreover, our work deeply investigated the gap between affective content analysis and induced emotion recognition by gaining insight into the relationships between aesthetic highlights, induced emotions, and perceived emotions.

This paper describes work in progress to answer the question of how we can identify and model the depth and quality of student participation in class discussion forums using the content of the discussion forum messages. We look at two widely-studied frameworks for assessing critical discourse and cognitive engagement: the ICAP and Community of Inquiry (CoI) frameworks. Our goal is to discover where they agree and where they offer complementary perspectives on learning.

Being able to detect topics and speaker stances in conversations is a key requirement for developing spoken language understanding systems that are personalized and adaptive. In this work, we explore how topic-oriented speaker stance is expressed in conversational speech. To do this, we present a new set of topic and stance annotations of the CallHome corpus of spontaneous dialogues. Specifically, we focus on six stances-positivity, certainty, surprise, amusement, interest, and comfort-which are useful for characterizing important aspects of a conversation, such as whether a conversation is going well or not. Based on this, we investigate the use of neural network models for automatically detecting speaker stance from speech in multi-turn, multi-speaker contexts. In particular, we examine how performance changes depending on how input feature representations are constructed and how this is related to dialogue structure. Our experiments show that incorporating both lexical and acoustic features is beneficial for stance detection. However, we observe variation in whether using hierarchical models for encoding lexical and acoustic information improves performance, suggesting that some aspects of speaker stance are expressed more locally than others. Overall, our findings highlight the importance of modelling interaction dynamics and non-lexical content for stance detection.

The widespread use of online discussion forums in educational settings provides a rich source of data for researchers interested in how collaboration and interaction can foster effective learning. Such online behaviour can be understood through the Community of Inquiry framework, and the cognitive presence construct in particular can be used to characterise the depth of a student's critical engagement with course material. Automated methods have been developed to support this task, but many studies used small data sets, and there have been few replication studies.

Predicting the emotional response of movie audiences to affective movie content is a challenging task in affective computing. Previous work has focused on using audiovisual movie content to predict movie induced emotions. However, the relationship between the audience's perceptions of the affective movie content (perceived emotions) and the emotions evoked in the audience (induced emotions) remains unexplored. In this work, we address the relationship between perceived and induced emotions in movies, and identify features and modelling approaches effective for predicting movie induced emotions. First, we extend the LIRIS-ACCEDE database by annotating perceived emotions in a crowd-sourced manner, and find that perceived and induced emotions are not always consistent. Second, we show that dialogue events and aesthetic highlights are effective predictors of movie induced emotions. In addition to movie based features, we also study physiological and behavioural measurements of audiences. Our experiments show that induced emotion recognition can benefit from including temporal context and from including multimodal information. Our study bridges the gap between affective content analysis and induced emotion prediction.

As long-form spoken documents become more ubiquitous in everyday life, so does the need for automatic discourse segmentation in spoken language processing tasks. Although previous work has focused on broad topic segmentation, detection of finer-grained discourse units, such as paragraphs, is highly desirable for presenting and analyzing spoken content. To better understand how different aspects of speech cue these subtle discourse transitions, we investigate automatic paragraph segmentation of TED talks. We build lexical and prosodic paragraph segmenters using Support Vector Machines, AdaBoost, and Long Short Term Memory (LSTM) recurrent neural networks. In general, we find that induced cue words and supra-sentential prosodic features outperform features based on topical coherence, syntactic form and complexity. However, our best performance is achieved by combining a wide range of individually weak lexical and prosodic features, with the sequence modelling LSTM generally outperforming the other classifiers by a large margin. Moreover, we find that models that allow lower level interactions between different feature types produce better results than treating lexical and prosodic contributions as separate, independent information sources.

We present the Beetle-Grow intelligent tutoring system, which combines active experimentation, self-explanation, and formative feedback using natural language interaction. It runs in a standard web browser and has a fresh, engaging design. The underlying back-end system has previously been shown to be highly effective in teaching basic electricity and electronics concepts. Beetle-Grow has been designed to capture student interaction and indicators of learning in a form suitable for data mining, and to support future work on building tools for interactive tutoring that improve after experiencing interaction with students, as human tutors do.

We will demonstrate the Beetle-Grow intelligent tutoring system, which combines active experimentation, self-explanation, and formative feedback using natural language interaction. It runs in a standard web browser and has a fresh, engaging design. The underlying back-end system has previously been shown to be highly effective in teaching basic electricity and electronics concepts.

Automatic emotion recognition is vital for building natural and engaging human-computer interaction systems. Combining information from multiple modalities typically improves emotion recognition performance. In previous work, features from different modalities have generally been fused at the same level with two types of fusion strategies: Feature-Level fusion, which concatenates feature sets before recognition; and Decision-Level fusion, which makes the final decision based on outputs of the unimodal models. However, different features may describe data at different time scales or have different levels of abstraction. Cognitive Science research also indicates that when perceiving emotions, humans use information from different modalities at different cognitive levels and time steps. Therefore, we propose a Hierarchical fusion strategy for multimodal emotion recognition, which incorporates global or more abstract features at higher levels of its knowledge-inspired structure. We build multimodal emotion recognition models combining state-of-the-art acoustic and lexical features to study the performance of the proposed Hierarchical fusion. Experiments on two emotion databases of spoken dialogue show that this fusion strategy consistently outperforms both Feature-Level and Decision-Level fusion. The multimodal emotion recognition models using the Hierarchical fusion strategy achieved state-of-the-art performance on recognizing emotions in both spontaneous and acted dialogue.

In this work, we compare emotion recognition on two types of speech: spontaneous and acted dialogues. Experiments were conducted on the AVEC2012 database of spontaneous dialogues and the IEMOCAP database of acted dialogues. We studied the performance of two types of acoustic features for emotion recognition: knowledge-inspired disfluency and nonverbal vocalisation (DIS-NV) features, and statistical Low-Level Descriptor (LLD) based features. Both Support Vector Machines (SVM) and Long Short-Term Memory Recurrent Neural Networks (LSTM-RNN) were built using each feature set on each emotional database. Our work aims to identify aspects of the data that constrain the effectiveness of models and features. Our results show that the performance of different types of features and models is influenced by the type of dialogue and the amount of training data. Because DIS-NVs are less frequent in acted dialogues than in spontaneous dialogues, the DIS-NV features perform better than the LLD features when recognizing emotions in spontaneous dialogues, but not in acted dialogues. The LSTM-RNN model gives better performance than the SVM model when there is enough training data, but the complex structure of a LSTM-RNN model may limit its performance when there is less training data available, and may also risk over-fitting. Additionally, we find that long distance contexts may be more useful when performing emotion recognition at the word level than at the utterance level.

Automatic emotion recognition has long been a focus of Affective Computing. We aim at improving the performance of state-of-the-art emotion recognition in dialogues using novel knowledge-inspired features and modality fusion strategies. We propose features based on disfluencies and nonverbal vocalisations (DIS-NVs), and show that they are highly predictive for recognizing emotions in spontaneous dialogues. We also propose the hierarchical fusion strategy as an alternative to current feature-level and decision-level fusion. This fusion strategy combines features from different modalities at different layers in a hierarchical structure. It is expected to overcome limitations of feature-level and decision-level fusion by including knowledge on modality differences, while preserving information of each modality.

Papers on voice interfaces for people with cognitive impairment or demenita only provide small snapshots of actual interactions, if at all. This is a major obstacle to the development of better interfaces. Transcripts of interactions between users and systems contain rich evidence of typical language patterns, indicate how users conceptualise their computer interlocutor, and highlight key design issues. In this paper, we introduce the CADENCE corpus and outline how it can be used to stimulate replicable research on inclusive voice interfaces. The CADENCE corpus is first data set of its kind to include rich data from people with cognitive impairment and free for research use. The corpus consists of transcribed spoken interactions between older people with and without cognitive impairment and a simulated Intelligent Cognitive Assistant and includes comprehensive data on users' cognitive abilities.

In this paper, we show how a large demographic data set that includes only high-level information about health and disability can be used to specify user requirements for people with specific needs and impairments. As a case study, we consider adapting spoken dialogue systems (SDS) to the needs of older adults. Such interfaces are becoming increasingly prevalent in telecare and home care, where they will often be used by older adults.

In spoken dialog systems, information must be presented sequentially, making it difficult to quickly browse through a large number of options. Recent studies have shown that user satisfaction is negatively correlated with dialog duration, suggesting that systems should be designed to maximize the efficiency of the interactions. Analysis of the logs of 2,000 dialogs between users and nine different dialog systems reveals that a large percentage of the time is spent on the information presentation phase, thus there is potentially a large pay-off to be gained from optimizing information presentation in spoken dialog systems. This article proposes a method that improves the efficiency of coping with large numbers of diverse options by selecting options and then structuring them based on a model of the user's preferences. This enables the dialog system to automatically determine trade-offs between alternative options that are relevant to the user and present these trade-offs explicitly. Multiple attractive options are thereby structured such that the user can gradually refine her request to find the optimal trade-off. To evaluate and challenge our approach, we conducted a series of experiments that test the effectiveness of the proposed strategy. Experimental results show that basing the content structuring and content selection process on a user model increases the efficiency and effectiveness of the user's interaction. Users complete their tasks more successfully and more quickly. Furthermore, user surveys revealed that participants found that the user-model based system presents complex trade-offs understandably and increases overall user satisfaction. The experiments also indicate that presenting users with a brief overview of options that do not fit their requirements significantly improves the user's overview of available options, also making them feel more confident in having been presented with all relevant options.

In this paper, we show that the performance measures Pk and Window Diff, commonly used for discourse, topic, and story segmentation evaluation, are biased in favor of segmentations with fewer or adjacent segment boundaries. By analytical and empirical means, we show how this results in a failure to penalize substantially defective segmentations. Our novel unbiased measure k-κ corrects this, providing a single score that accounts for chance agreement. We also propose additional statistics that may be used to characterize important properties of segmentations such as boundary clumping. We go on to replicate a recent spoken-language topic segmentation experiment, drawing conclusions that are substantially different from previous studies concerning the effectiveness of state-of-the-art topic segmentation algorithms.

Generating responses that take user preferences into account requires adaptation at all levels of the generation process. This article describes a multi-level approach to presenting user-tailored information in spoken dialogues which brings together for the first time multi-attribute decision models, strategic content planning, surface realization that incorporates prosody prediction, and unit selection synthesis that takes the resulting prosodic structure into account. The system selects the most important options to mention and the attributes that are most relevant to choosing between them, based on the user model. Multiple options are selected when each offers a compelling trade-off. To convey these trade-offs, the system employs a novel presentation strategy which straightforwardly lends itself to the determination of information structure, as well as the contents of referring expressions. During surface realization, the prosodic structure is derived from the information structure using Combinatory Categorial Grammar in a way that allows phrase boundaries to be determined in a flexible, data-driven fashion. This approach to choosing pitch accents and edge tones is shown to yield prosodic structures with significantly higher acceptability than baseline prosody prediction models in an expert evaluation. These prosodic structures are then shown to enable perceptibly more natural synthesis using a unit selection voice that aims to produce the target tunes, in comparison to two baseline synthetic voices. An expert evaluation and f0 analysis confirm the superiority of the generator-driven intonation and its contribution to listeners' ratings.

We propose that a set of five dimensions forms a foundation underlying a number of prevalent theoretical perspectives on learning. We show how student contributions to instructional dialogue can be reliably annotated with these dimensions. Finally, we provide preliminary validation evidence for our coding scheme and illustrate the potential value of such an approach to analyzing student behavior in tutorial dialogue.

Our research goal is to investigate whether previous findings and methods in the area of tutorial dialogue can be generalized across dialogue corpora that differ in domain (mechanics versus electricity in physics), modality (spoken versus typed), and tutor type (computer versus human). We first present methods for unifying our prior coding and analysis methods. We then show that many of our prior findings regarding student dialogue behaviors and learning not only generalize across corpora, but that our methodology yields additional new findings. Finally, we show that natural language processing can be used to automate some of these analyses.

Advances in multimedia technologies have enabled the creation of huge archives of audio-video recordings of meetings, and there is burgeoning interest in developing meeting browsers to help users better leverage these archives. A recent study has shown that extractive summaries provide a more efficient way of navigating meeting content than simply reading through the transcript and using the audio-video record, or navigating via keyword search (Murray, 2007). The extractive summary technique identifies informative dialogue acts to generate general purpose summaries. These summaries can still be lengthy. Recently, we have developed a decision-focused summarization system that presents only 1-2% of the recordings related to decision making. In this paper, we describe a task-based evaluation in which we compare the decision-focused summaries to the general purpose summaries. Our results indicate that the more focused summaries help users perform the decision debriefing task more effectively and improve perceived efficiency. In addition, this study also investigates the effect of automatic summaries and transcription on task effectiveness, report quality, and users' perceptions of task success.

We evaluated two strategies for alleviating working memory load for users of voice interfaces: presenting fewer options per turn and providing confirmations. Forty-eight users booked appointments using nine different dialogue systems, which varied in the number of options presented and the confirmation strategy used. Participants also performed four cognitive tests and rated the usability of each dialogue system on a standardised questionnaire. When systems presented more options per turn and avoided explicit confirmation subdialogues, both older and younger users booked appointments more quickly without compromising task success. Users with lower information processing speed were less likely to remember all relevant aspects of the appointment. Working memory span did not affect appointment recall. Older users were slightly less satisfied with the dialogue systems than younger users. We conclude that the number of options is less important than an accurate assessment of the actual cognitive demands of the task at hand.

Abstract Richly annotated dialogue corpora are essential for new research directions in statistical learning approaches to dialogue management, context-sensitive interpretation, and context-sensitive speech recognition. In particular, large dialogue corpora annotated with contextual information and speech acts are urgently required. We explore how existing dialogue corpora (usually consisting of utterance transcriptions) can be automatically processed to yield new corpora where dialogue context and speech acts are accurately represented. We present a conceptual and computational framework for generating such corpora. As an example, we present and evaluate an automatic annotation system which builds ‘Information State Update’ (ISU) representations of dialogue context for the Communicator (2000 and 2001) corpora of human–machine dialogues (2,331 dialogues). The purposes of this annotation are to generate corpora for reinforcement learning of dialogue policies, for building user simulations, for evaluating different dialogue strategies against a baseline, and for training models for context-dependent interpretation and speech recognition. The automatic annotation system parses system and user utterances into speech acts and builds up sequences of dialogue context representations using an ISU dialogue manager. We present the architecture of the automatic annotation system and a detailed example to illustrate how the system components interact to produce the annotations. We also evaluate the annotations, with respect to the task completion metrics of the original corpus and in comparison to hand-annotated data and annotations produced by a baseline automatic system. The automatic annotations perform well and largely outperform the baseline automatic annotations in all measures. The resulting annotated corpus has been used to train high-quality user simulations and to learn successful dialogue strategies. The final corpus will be made publicly available.

A 2x2 enhanced Wizard-of-Oz experiment (N = 32) was conducted to compare two different approaches to presenting information to drivers in easy and difficult driving conditions. Data of driving safety, evaluation of the spoken dialogue system, and perception of self were analyzed. Results show that the user-modeled summarize-and-refine (UMSR) approach led to more efficient information retrieval than did the summarize-and-refine (SR) approach. However, depending on driving condition, higher efficiency did not always translate into pleasant subjective experience. Implications for usability testing and interface design were presented, followed by discussions of future research directions.

This paper presents results of a Wizard-of-Oz (WoZ) study examining the effect of two different information presentation methods on a secondary task, namely driving. The results demonstrate that the user-model based summarize and refine (UMSR) approach enables more efficient information retrieval in comparison to the data-driven summarize and refine (SR) approach, and does not negatively affect driving performance.

We report the results of a Wizard-of-Oz (WoZ) study comparing two approaches to presenting information in a spoken dialogue system generating flight recommendations. We found that recommendations presented using the user-model based summarize and refine (UMSR) approach enable more efficient information retrieval than the data-driven summarize and refine (SR) approach. In addition, user ratings on four evaluation criteria showed a clear preference for recommendations based on the UMSR approach.

In this paper we present how the automatic extraction of events from text can be used to both classify narrative texts according to plot quality and produce advice in an interactive learning environment intended to help students with story writing. We focus on the story rewriting task, in which an exemplar story is read to the students and the students rewrite the story in their own words. The system automatically extracts events from the raw text, formalized as a sequence of temporally ordered predicate-arguments. These events are given to a machine-learner that produces a coarse-grained rating of the story. The results of the machine-learner and the extracted events are then used to generate fine-grained advice for the students.

This study concerns how to segment a scenario-driven multiparty dialogue and how to label these segments automatically. We apply approaches that have been proposed for identifying topic boundaries at a coarser level to the problem of identifying agenda-based topic boundaries in scenario-based meetings. We also develop conditional models to classify segments into topic classes. Experiments in topic segmentation show that a supervised classification approach that combines lexical and conversational features outperforms the unsupervised lexical chain-based approach, achieving 20% and 12% improvement on segmentating top-level and sub-topic segments respectively. Experiments in topic classification suggest that it is possible to automatically categorize segments into appropriate topic classes given only the transcripts. Training with features selected using the Log Likelihood ratio improves the results by 13.3%.

Clarification requests (CRs) in conversation ensure and maintain mutual understanding and thus play a crucial role in robust dialogue interaction. In this paper, we describe a corpus study of CRs in task-oriented dialogue and compare our findings to those reported in two prior studies. We find that CR behavior in task-oriented dialogue differs significantly from that in everyday conversation in a number of ways. Moreover, the dialogue type, the modality and the channel quality all influence the decision of when to clarify and at which level of the grounding process. Finally we identify form-function correlations which can inform the generation of CRs.

We have developed a system that generates evaluative arguments that are tailored to the user, properly arranged and concise. We have also developed an evaluation framework in which the effectiveness of evaluative arguments can be measured with real users. This paper presents the results of a formal experiment we have performed in our framework to verify the influence of argument conciseness on argument effectiveness

We propose an argumentation strategy for generating evaluative arguments that can be applied in systems serving as personal assistants or advisors. By following guidelines from argumentation theory and by employing a quantitative model of the user's preferences, the strategy generates arguments that are tailored to the user, properly arranged and concise. Our proposal extends the scope of previous approaches both in terms of types of arguments generated, and in terms of compliance with principles from argumentation theory.

Our goal is to identify the features that predict the occurrence and placement of discourse cues in tutorial explanations in order to aid in the automatic generation of explanations. Previous attempts to devise rules for text generation were based on intuition or small numbers of constructed examples. We apply a machine learning program, C4.5, to induce decision trees for cue occurrence and placement from a corpus of data coded for a variety of features previously thought to affect cue usage. Our experiments enable us to identify the features with most predictive power, and show that machine learning can be used to induce decision trees useful for text generation.

We claim that automatic multimedia presentation can be modeled by integrating two complementary approaches to automatic design: hierarchical planning to achieve communicative goals, and task-based graphic design. The interface between the two approaches is a domain and media independent layer of communicative goals and actions. A planning process decomposes domain-specific goals to domain-independent goals, which in turn are realized by media-specific techniques. One of these techniques is taskbased graphic design. We apply our approach to presenting information from large data sets using natural language and information graphics. Keywords Multimedia presentation, information seeking tasks, media allocation, information graphics, presentation planning. INTRODUCTION Understanding large data sets and explaining them to others via effective displays is a complex, laborious and time consuming activity. Analysts and other types of specialists on a daily and sometimes hourly basis explor...

Our goal is to identify the features that predict cue selection and placement in order to devise strategies for automatic text generation. Much previous work in this area has relied on ad hoc methods. Our coding scheme for the exhaustive analysis of discourse allows a systematic evaluation and refinement of hypotheses concerning cues. We report two results based on this analysis: a comparison of the distribution of SINCE and BECAUSE in our corpus, and the impact of embeddedness on cue selection.

Most complex systems provide some form of help facilities. However, typically, such help facilities do not allow users to ask follow up questions or request further elaborations when they are not satisfied with the systems' initial offering. One approach to alleviating this problem is to present the user with a menu of possible follow up questions at every point. Limiting follow up information requests to choices in a menu has many advantages, but there are also a number of issues that must be dealt with in designing such a system. To dynamically generate useful embedded menus, the system must be able to, among other things, determine the context of the request, represent and reason about the explanations presented to the user, and limit the number of choices presented in the menu. This paper discusses such issues in the context of a patient education system that generates a natural language description in which the text is directly manipulable - clicking on portions of the text causes the system to generate menus that can be used to request elaborations and further information.

When participating in tutorial dkdogues, human tutors freely refer to their own previous explanations.Explanation is an inherently incremental and interactive process.New information must be highlighted and related to what has already been presented.If user interfaces are to reap the benefits of natural language interaction, they must be endowed with the properties that make human natural language interaction so effective.TMS paper describes the design of a user interface that enables both the system and the user to refer to the past dialogue.The work is based on the notion that the dialogue history is a source of knowledge that can be manipulated like any other.In particukac, we describe an interface that allows students to visualize the dklogue history on the screen, highlight its relevant parts and query and manipulate the dialogue history.We expect that these facilities will increase the effectiveness of the student learning of the task.

Explanation is an interactive process requiring a dialogue between advice-giver and advice-seeker. In this paper, we argue that in order to participate in a dialogue with its users, a generation system must be capable of reasoning about its own utterances and therefore must maintain a rich representation of the responses it produces. We present a text planner that constructs a detailed text plan, containing the intentional, attentional, and rhetorical structures of the text it generates.

Expert and advice-giving systems produce complex multi-sentential responses to user's queries. Results from analyses of novice/expert dialogues indicate that novices often do not understand an expert's response and rarely ask a well-formulated follow-up question. Thus systems must be able to provide further information in response to vaguely articulated questions. However, current systems cannot clarify misunderstood explanations or elaborate on previous explanations. In this paper we describe an approach to explanation generation that expands a system's explanatory capabilities and enables the production of clarifying or elaborating explanations in response to follow-up questions or indication that the explanation was not understood.

Principled development techniques could greatly enhance the understandability of expert systems for both users and system developers. Current systems have limited explanatory capabilities and present maintenance problems because of a failure to explicitly represent the knowledge and reasoning that went into their design. This paper describes a paradigm for constructing expert systems which attempts to identify that tacit knowledge, provide means for capturing it in the knowledge bases of expert systems, and, apply it towards more perspicuous machine-generated explanations and more consistent and maintainable system organization.

Atomic transactions are useful in distributed systems as a means of providing reliable operation in the face of hardware failures. Nested transactions are a generalization of traditional transactions in which transactions may be composed of other transactions. The programmer may initiate several transactions from within a transaction, and serializability of the transactions is guaranteed even if they are executed concurrently. In addition, transactions invoked from within a given transaction fail independently of their invoking transaction and of one another, allowing use of alternate transactions to accomplish the desired task in the event that the original should fail. Thus nested transactions are the basis for a general-purpose reliable programming environment in which transactions are modules which may be composed freely.