SARA- Socially Aware Robot Assistant Assignment

Question

Task: For this component you will be required to do a 5-10 minute presentation on a recent academic paper on a topic related to Intelligent Systems for Analytics or Intelligent Systems. Some possible topic areas include but are not limited to:

• Intelligent Systems for Data Warehouse systems
• Evolving Intelligent Systems: Methods, Learning, & Applications
• Distance Metric Learning in Intelligent Systems
• Intelligent Systems for Socially Aware Computing
• Data Mining techniques with IS
• frameworks for integrating AI and data mining
• Expert System
• Structure of knowledge Engineering
• IS and Support Vector Machines
• IS and Neural Network Architectures
• Heuristic Search Methods
• Genetic Algorithms and Developing GA Applications

The paper you select must be directly relevant to one of the above topics or another topic and be related to Intelligent Systems for Analytics. The paper must be approved by your lecturer and be related to what we are studying this semester in Intelligent Systems for Analytics. The paper MITS5509 Assignment 1 and 2 Copyright © 2015-2019 VIT, All Rights Reserved. 3 can be from any academic conference or other relevant Journal or online sources such as Google Scholar, Academic department repositories, or a significant commercial company involved in research such as IBM etc. All students must select a different paper. Thus, the paper must be approved by your lecturer before proceeding. In case two students are wanting to present on the same paper, the first who emails the lecturer with their choice will be allocated that paper. Please note that popular magazine or web-site articles are not academic papers.

A grade of 10% of the Units mark will be awarded for your presentation and your participation in other student presentations. You are to prepare a set of powerpoint slidesfor your presentation. If you do not participate in at least 70% of other student’s presentations you will forfeit a significant proportion of the marks for this component.

Answer

Introduction
Socially Aware Robot Assistant (SARA) discussed in this socially aware robot assistant assignment can be used to analyze the visual movement such as head movement and face action of the user along with his acoustic feature (vocal) and communicational strategies (verbal) behaviours [1]. The technological process helps to identify the user’s rapport level regarding these movements. The program also helps to achieve social tasks along with task goals by using its personal verbal, visual and vocal behaviours. Rapport can be made by the technological tool by eliciting the preferences of the attendees through the presentation of the agent aids. After that, personalized recommendations are being informed about the existing sessions for attending and meeting the people. Hence, it has been identified that applications such as Amazon Alexa, Microsoft Cortana and especially Apple Siri have evaluated the factors of personal assistants regarding technological improvement.

The following assignment is based on SARA’s computational architecture along with its impact on the technological world. How Sara engages the conversation level by an animated character and how today world has adopted the approach would also be discussed in the study.

Sara’s computational architecture
SARA or Socially Aware Robot Assistant has been designed for building an interpersonal and specific closeness of a clear conversation for the purpose of improving the level of understanding and the user’s visual, verbal and vocal behaviour. The rapport’s dynamic has led the current work to be prior along with a rapport building agent’s computational architecture for the activity’s initial consideration [3]. The architectural overview of the SARA’s computing system has helped the current technology system to be improved, and it engages a significant number of customers in terms of evolving the business in the competitive market. The computing system’s entire modules have been built upon the Virtual Human Toolkit’s top, and it can be said the most responsible equipment behind the successful representation of SARA’s computational architecture. Architectural modules that are responsible for creating SARA are described below:

1.  Visual and vocal input analysis in socially aware robot assistant assignment: The texts are being converted to the speech through the help of the API system, which is a cognitive service of the Microsoft system. Identification of the users has been made by LUIS ( Language Understanding Intelligent Service) software, which is also a Microsoft system). The specified domain’s data has been identified as limited that it cannot be able to train the domain data. Oz GUI’s wizard has been created as a supporting backup in case of emergency in the module of visual and vocal input analysis regarding the identification of the speech and error regarding the natural language. OpenSmile has been implemented for extracting the acoustic features and services from the audio signal including SMA (loudness), FO (Fundamental Frequency) and moreover, jitter and shimmer [2]. Jitter and Shimmer are used for input serving to the rapport estimate system, and therefore, the classifier modules of the conversational strategy are being served. 3D facial landmarks, gaze, action units along with head pose are being detected by the Openface application for detecting the visual and vocal input analysis regarding the identification of the speech and error regarding the natural language. These factors are also being used for serving to the rapport estimate system as an input, and it has a positive impact on technological improvement [5].
2. conversational strategy classifier: A multimodal classifier along with the conversational strategy has been implemented in the SARA (Socially Aware Robot Assistant) for the purpose of recognizing some specific styles and particular strategies of the conversational style for building along with maintaining the budding relationship and sometimes it can be destroyed by the conversational style. The methods include eliciting self-disclosure (QE), praise (PR), violation of social norms (VSN) and reference to shared experience (RSD) [4]. The dialogues can be successfully recognized through the analysis of the rich and contextual features. It has been drawn from the speaker’s vocal, verbal and visual modalities along with the activities of the interlocutor regarding both the previous and the current turns. An appropriate accuracy of user input has been identified as more than 80%, and the identified Kappa’s accuracy is more than 60%.
3. Rapport estimator: An automatic and appropriate rapport estimator has been implemented in the SARA or Socially Aware Robot Assistant system. It has been based on the temporal association’s rule learning framework. It has been implemented for performing a specific and appropriate investigation. How the interlocutor’s behaviours sequences have been led for improving and degrading the impacts in the interpersonal rapport system of conversation also have been discussed. The analysis of the visual behaviours discussed in this socially aware robot assistant assignment includes smiles of the users along with eye gaze and the strategies of verbal communication in the interpersonal rapport system such as VSN, PR, SD, BC and RSE [7]. The learned rules of the temporal association involve two-step fusion in the Rapport forecasting model. The first step deals with the goals that aim to learn each temporal association’s contribution to the current report’s presence or the absence through the classifiers of the seven random forest. The second step is all about the learning of each binary classifier regarding the corresponding for the prediction of the continuous rapport value.  The peer tutoring sessions have been divided into 30 seconds of videos for the identification of the ground truth, and thereafter, the videos are randomized. 
4. Dialogue management: The essentiality of dialogue management is to composing a task reasoner that would help to focus on the information that is obtained for fulfilling the goals and objectives of the user. Along with that, it also helps to choose the way how a social reasoner would talk to build rapport in the system of the service for a better achievement of the user’s goals and objectives. The task and its social history, along with a user model, play an essential role in dialogue management discussed in this socially aware robot assistant assignment.
   • Task reasoner: Task reasoner plays a crucial role in maintaining the system initiative to the possible extent. It has been implemented in the SARA or Socially Aware Robot Assistant for identifying the finite state machine. Several triggering events’ transitions can determine the user's intents such as the dialogue's past and present state along with some contextual information by the implementation of a task reasoner.
   • 2.4.2 Social Reasoner: The transition of this machine has been determined by several nodes trig sequences such as VSN, PR, SD, BC and RSE. The system is essential for establishing a respectful along with a cordial communication system with the respected user along with the utilization of SD, which is an ice-breaking strategy. It has been followed by PR to encourage the respected user. It also helps to perform SD. after that, and a VSN is being performed after the proper interaction along with the high rapport level. It has been identified that the output of the task reasoner can be either as a query form of the main domain database or an intended system. It would help to serve the social reasoner’s input system. The NLG modules would also be used in this study. The user’s request could be handled by this system, along with a taken approach to deal with the errors. The Alex framework’s implementation has helped to acknowledge it [1].
5. NLG and behaviour generation: This system described in socially aware robot assistant assignment depends on the dialogue manager’s output that includes the system intent, expected conversational strategy and the present conversational state for improving the quality of SARA. The behaviours and attitudes of the users can be generated through his system. The NLG or National Language Generator has helped to select the syntactic templates that are associated with some specific conversational strategies to perform the user’s queries. BEAT has been implemented in this system for generating the non-verbal behaviour of the users. Virtual Human toolkit has also been a part of this system for helping the output to be generated and identified. After that, the plan would be sent to the Smartbody for rendering the essential non-verbal behaviours [5].

Three dialogue examples
It has been identified that SARA or Socially Aware Robot Assistant was discovered in Tianjin China at the World Economic Forum in June of 2016. It was served to the specific event app as a front end app. The interaction with SARA was with almost 100 participants for getting advice and meeting the people. Oz GUI’s wizard has been created as a supporting backup in case of emergency in the module of visual and vocal input analysis regarding the identification of the speech and error regarding the natural language. OpenSmile has been implemented for extracting the acoustic features and services from the audio signal including SMA (loudness), FO (Fundamental Frequency) and moreover, jitter and shimmer [6].

Conclusion
After analysing the context of socially aware robot assistant assignment, it can be concluded that SARA stands for Socially Aware Robot Assistant which can be used to analyze the visual movement such as head movement and face movement of the user along with his acoustic feature (vocal) and communicational strategies (verbal) behaviours. SARA or Socially Aware Robot Assistant had been designed for building an interpersonal and specific closeness of a clear conversation for the purpose of improving the level of understanding and the user’s visual, verbal and vocal behaviour. The transition of this machine has been determined by several nodes trig sequences such as VSN, PR, SD, BC and RSE.

Bibliography:
[1] Zetsche, Bernd, Sara E. Volz, and Feng Zhang. "A split-Cas9 architecture for inducible genome editing and transcription modulation." Nature biotechnology 33.2 (2015): 139.

[2] Morley-Short, Sam, et al. "Physical-depth architectural requirements for generating universal photonic cluster states." Quantum Science and Technology 3.1 (2017): 015005.

[3] Braun, Daniel, Ehud Reiter, and Advaith Siddharthan. "SaferDrive: An NLG-based behaviour change support system for drivers." Natural Language Engineering 24.4 (2018): 551-588.

[4] Foster, Mary Ellen. socially aware robot assistant assignment "Natural language generation for social robotics: opportunities and challenges." Philosophical Transactions of the Royal Society B 374.1771 (2019): 20180027.

[5] Su, Pei-Hao, et al. "Continuously learning neural dialogue management." arXiv preprint arXiv:1606.02689 (2016).

[6] Cuayáhuitl, Heriberto, Simon Keizer, and Oliver Lemon. "Strategic dialogue management via deep reinforcement learning." arXiv preprint arXiv:1511.08099 (2015).

[7]Gkatzia, Dimitra, Oliver Lemon, and Verena Rieser. "Natural language generation enhances human decision-making with uncertain information." arXiv preprint arXiv:1606.03254(2016).