A Review Of Conversation Interface Technology In Education And Training – Essay Example

A Review of Conversation Interface Technology in Education and Training Conversational user interfaces (CUIs) allow machines tounderstand various natural communication modes such that even with the increasing complexity of technology, it shall enable human-computer communication to remain easy for all to learn and use. This means that CUIs will be at the forefront in helping man discover better solutions to multiple domains. It is believed that CUIs are the technology that will move us away from this Information Age to a new Age. There is no sector where CUIs have begun to play as big a role as they have in that for Education and Training. This paper introduces us to this new technology, gives us an example of a working CUI, informs us about the greater challenges that it currently faces, and discusses why this technology is ripe for the Education and Training sector.
Keywords: Conversational user interfaces, dialogue systems, Human Language Technologies (HLTs), e-learning, Natural Language Processing (NLP), Computer Supported Collaborative Learning (CSCL), interactive learning, education and training
Man’s continuous advancement through the years has been characterized by certain major innovations that have led him from the Agrarian Age to the Industrial Age up to the Information Age where we are. The invention of the digital computer is believed to have been the turning point of the 20th century and the facilitator for the move from the Industrial Age to the Information Age. We have been in the Information Age for quite a while and with the rapid technological developments, many scientists are of the opinion that man will soon usher in a new era that [1] refers to as the Symbiotic Age. In a manner similar to all the other transitions through the Ages, technological innovation shall play a huge role here as well. [1] and others believe that the emergence of the Conversational User Interface (CUI) on the global web marks the beginning of the coming of a new Age especially when one considers man’s increasing inseparable connection to technological infrastructure. [1] goes further to state that there is good early evidence as to why CUIs are the revolutionary technology. He says CUIs will help man discover better collective solutions in multiple domains such as globalization, education, governance, environment, and health among others. This will happen because conversational user interfaces will enhance our capacity to extract knowledge from our already large and yet rapidly increasing databases of information [1].
Natural Language Processing
The whole idea behind conversational interfaces is to copy human-human interactions and the most attractive means of communication for humans is the spoken language. Human speech is flexible, efficient, inexpensive and ideal for communication, however when we view it from the machine point of view we find “human dialogues to be variable, containing frequent interruptions, speech overlaps, incomplete or unclear sentences, incoherent segments, and topic switches” [4]. Human communication is multimodal and even that which we would expect to be simpler, verbal communication, involves multiple direct and indirect activities.
Natural Language Processing is the field that studies the interaction between human and computer languages. Natural Language Processing aids in the development of computer systems that comprehend and manipulate natural languages. [11] suggests that the ultimate purpose of Natural Language Processing is to achieve human-like language processing to facilitate human-machine interaction. Natural Language Processing (NLP) has two branches: language processing and language generation. According to [11] language processing looks at the analysis of language for the purpose of producing a meaningful representation. From such representations we are then able to produce a language which is what NLP language generation is all about.
The process of building computer applications that comprehend natural language has three main problems: the first one relates to the thought process, the second one to the representation and meaning of the linguistic input, and the third one to the world knowledge. Applications of NLP can be found both in theory and in implementation such as information retrieval and extraction, machine translation, natural language text processing and summarization, multilingual and cross language information retrieval (CLIR), artificial intelligence, dialogue systems, and so on.
Conversational User Interfaces
Self-service systems, online help systems, web services, mobile communication devices, remote control systems, and dashboard computers are providing ever more functionality which implies greater complexity and a steeper learning curve for users. Conversational user interfaces allow various natural communication modes such as gestures and speech for input and output while exploiting the context in which an input is used to compute its meaning. The increasing interest in CUIs is driven by our desire to support natural, flexible, efficient and powerfully expressive means of human-computer communication that shall be easy for all to learn and use [2]. [2] adds that the development of conversational user interfaces will enable the average person to interact with computers anytime and anywhere without special skills or training, using such common devices as a mobile phone.
It is vital to note that CUIs differ generally in the degree to which the system or user controls the conversation. Currently there are three broad categories: directed dialogue, ‘free-form’ dialogue and mixed dialogue systems. According to [3]:
The categorization of these spoken dialogue systems is differentiated by the degree with which the system maintains control of the conversation, and the inherent amount of flexibility provided to the user to ask for a) what they want, b) in the way they want to ask for it, and c) when they want to ask it (p.350).
In directed dialogue systems the user is restricted in terms of option, similar to the interactive voice response systems that we encounter often in phone-in customer care centers. In directed dialogue systems users have a set of prescribed functions at their disposal. It is easy to develop systems with such a restricted framework and that is why directed dialogue systems are the first conversational user interfaces to be deployed on a wide scale, successfully [3]. Free-form dialogue systems are the exact opposite of directed dialogue systems because complete control is maintained by humans while the system remains passive.
In mixed dialogue CUIs, the approach employed offers a more flexible dialogue strategy that empowers both the user and the machine such that dialogue resembles the human face-to-face conversations where both parties have control [3]. Therefore, it is no surprise that systems built to operate in this paradigm will be typically more complex than their directed dialogue counterparts [4], and hence more difficult to design and deploy. For this reason, most mixed-initiative systems remain under development in research laboratories.
Although mixed-initiative CUIs are the ultimate natural and efficient means of communication, their widespread use has been hampered by multiple fundamental technical barriers. The greatest of these technical barriers is the predicament of configuring Human Language Technologies (HLTs) necessary for the creation of mixed-initiative prototype systems. At the moment, most of the development in taking place is language specific and domain-dependent. This has meant that dialogue management must be fine-tuned differently for each application. This application-specificity eliminates the portability of these Human Language Technologies across different acoustic environments, databases, and knowledge domains [3]. For conversational interfaces to become as ubiquitous as the telephone, researchers have no option but to seek ways to make it easier for developers to create systems that learn and improve their performance automatically.
In spite of the challenges facing the production of CUI systems researchers around the world continue to make inroads towards making the technology a common reality. For example researchers at Philip Research developed a system called Speech Interfaces for Consumer Electronics (SPICE). SPICE is a CUI to an Electronic Program Guide developed to support navigation in a large TV program database for controlling the TV set and also be used for programming a VCR. With SPICE users can interact with the TV by issuing commands through a touch-screen input on a hand-held GUI display in combination with spoken dialogue [5].
The main features of SPICE conversational interface that make it to become a powerful communication partner are: a natural language input, direct access to content, cooperative dialogue and choice and combination of modalities [5]. The devices we use to interact with out TV sets today have a pre-defined menu and prescribed keyword that we must remember in order to use the devices. With SPICE the natural language input enables users to simply state their request e.g. “What history shows are on the BBC1?” The direct access to content implies that when say one is searching for a movie in the TV database, we are no longer restricted by classifications such as genre, title etc. SPICE’s natural language capacity and large-vocabulary speech recognition makes it possible to retrieve the movie using unstructured input such as “Is there a James Bond movie today?” Cooperative dialogue in SPICE is made possible through facilitation of two-way communication between the user and the TV system. User’s input is complemented by the system as it offers suggestions based on the user’s preferences. Finally, SPICE allows users to choose the most suitable mode of input at that particular moment. The user could opt to use touch-screen input, speech or a combination of both [5].
Education and Training
As we have stated earlier, one of the greater impacts that shall be experienced from the emergence of a Conversational User Interface (CUI) on the global web is its benefits to education and training. A study by [6] in the world’s fastest growing economy, China, informs us about the rapid development of information and communication technology and its role in education. With its huge population, a drive towards e-learning in China would reflect a great increase in the audience for the adoption and use of conversational interface systems. By 2008, the number of universities, primary and secondary schools and scientific research institutes connecting to the China Education and Research Network (CERNET) had reached over 2000, with almost 30 million users. CERNET now has over 30 international and regional communication channels, which ensures safe and high-speed information exchange among educational institutions both at home and abroad [6].
The Chinese government, much like most western governments, is pushing for the steady development and reform of universities through establishment of e-campuses as a brand new model for teaching, conducting collaborative scientific research and management in colleges and universities. Our seemingly obsessive view with the occurrences in China emanate from the fact that its population represents one sixth of the total world population, therefore if its government can successfully encourage a shift towards online education and e-learning, then the demand for conversational interface systems would increase. A high demand implies more research funding and greater likelihood of success for the ubiquity of CUIs.
According to [6] the four characteristics of e-campus are: the digital storage of resources, information transfer via networks, automation of management and personalization of communication. It is within the objective for personalization of communication that we find the increasing interest in conversational user interfaces (CUIs). The development of hybrid and web-centric courses encourages a variety of interaction from the participants with the tutor fulfilling more of a facilitation role than a tutor [7].
In UK higher education establishments the focus has traditionally been very much on transferring a body of knowledge to a set of learners using a variety of teaching methods where conversation is only used as a means of clarifying the learning by the instructor [7]. With the advent of online teaching and learning, the UK’s Department for Education and Skills (DfES) sought to develop another approach. DfES funded the use of new technology to help raise academic standards through projects such as Talk 2 Learn. To help us comprehend the relevance of CUIs in education and training we shall give a brief description of Talk 2 Learn as an interactive online learning tool.
Talk 2 Learn began as a pilot project in 2000 with 1200 newly appointed UK head teachers. It organized people into various groupings known as communities so that members of each community negotiate meaning over a variety of professional issues and initiatives they are facing. Communities could be made to be private and membership ranges from one to thousands [7]. The Talk 2 Learn communities often do not follow a set of learning outcomes so as to allow for different viewpoints and perspectives and coming to a better understanding of issues through discussion and debate. The ultimate goal of the project was to encourage social interaction and individual participation in social behaviors such as learning. Vygotsky – a renowned scientist and expert in human cognitive development and interpersonal communication – considered that learning takes place more often outside a traditional setting, such as a school, than in it. One of the strengths of conversational interfaces is that they can be used to enhance learning outside the classroom through modern infrastructure such as mobile phones. According to [7]:
One of the distinguishing features of the Talk 2 Learn software from email listings and online discussion groups is its ability to organize communities in a wide variety of ways. It is very easy to set up new groupings and sub-groupings (p.251).
Currently, there are three types of online courses: those that present material with little or no interaction, web-enhanced courses consisting of a hybrid of face-to-face and online, and web-centric which are interactive courses conducted exclusively using a course site [8] It is in the latter interactive courses where the use of conversational user interfaces shall have the greatest impact. As Vygotsky’s concepts state, the makeup of learning groups and an emphasis on the interaction of learners with one another is a key to successful individual learning [7]. He goes further to state that in the past human beings formed communities that accumulated collective learning into social practices. In those communities knowledge was not an object; it was a living part of their practice. Knowing was an act of participation. Now that the Internet has become an essential tool assisting college students in learning it is necessary to develop tools that shall transform learning from being an object into part of our modern culture. The value of conversational interfaces to promote this shift today cannot be gainsaid.
Computer Supported Collaborative Learning (CSCL) was developed with the purpose being to provide opportunities to design and implement methods of advanced learning, such as deep learning, sustained and critical discourse, and effective discussion [9]. Current insights into the CSCL-environment suggest that social interaction is likewise important for communities that support learning [9]. Mixed-initiative CUIs enable web-centric courses to support more flexible dialogue strategies within the virtual classrooms by making the e-environment to be amiable for conversation. Group cohesion and social interaction are a necessary first step for collaborative learning or where effective discussion is expected to take place. [9] says that simply making a computer-supported discussion forum available does not guarantee that it will be used effectively to enable learning. We must find tools and methods that shall encourage conversation among all the parties involved (including the machines).
The current state of e-learning, however does not present us with a rosy outlook. In a study conducted by Rossett and Marshall in mid 2009 it was discovered that the direction of e-learning has not shifted much over the past several years. Most of e-learning conducted is about measuring and delivering through familiar instructional strategies such as tutorials and scenarios. Our expectation of increased personalized learning, problem solving and knowledge construction on the web remains low. The anticipated increase in online discussions to support knowledge transfer from the classroom or increasing use of mobile devices was unexpectedly rare [10]Use of Web 2.0 activities to promote user-generated content and collaboration remain scarce except in academia. In fact [10] shows that the most frequently occurring e-learning practice is still that of testing of skills and knowledge. From their research the major barriers to interactive, personalized e-learning were (in decreasing order): lack of financial resources, resistance to change and technology shortcomings.
It is almost certain that similar to the way we moved from the Industrial Age to the Information Age, we shall be moving into a newer Age. Our predictions at the moment inform us that in that coming Age the average human interaction with the average computer will not be through a mouse and a keyboard but through voice. This will be facilitated by the development and use of conversational user interfaces (CUIs). [1] predicts that CUIs will help us discover better collective solutions in several domains and that there is good early evidence to prove this. Among the greater solutions that we anticipate is the fall of political/power/equity divides due to the elimination of the current global inequity of access to high quality, lifelong education.
Conversational user interfaces allow various natural communication modes such as gestures and speech for input and output while exploiting the context in which an input is used to compute its meaning. This would lessen the need for users to continuously undergo steep learning curves whenever new technology is introduced. [2] believes that CUIs will enable the average person to interact with computers anytime and anywhere without special skills or training, using such common devices as a mobile phone.
[3] says that mixed-initiative CUIs are the ultimate natural and efficient means of communication because they employ a more flexible dialogue strategy that allows both the user and the machine to control dialogue much like how we conduct human face-to-face conversations. Within the field of education and training online and e-learning present the most obvious platform for the implementation of conversational user interfaces (CUIs). Moreover, more governments are willing to spend money towards the use of technology to raise academic performance. It is therefore imperative that we use this opportunity, when the Internet has become an essential tool assisting students in learning, to transform learning from being an object into part of our modern culture. Even though [10] demonstrates that we need to make substantial inroads in the arena of e-learning, as [1] states, a new Age is coming when the use of conversational interfaces shall be the norm. The future relations between man and computer shall be characterized by two possible scenarios: either the computers will be smart enough to communicate in human natural language, or humans will have to adapt their practices in order to communicate with computers [12].
1. Smart, J. M. The Conversational Interface: Our Next Great Leap Forward. Acceleration Watch, http://www.accelerationwatch.com/lui.html#interface
2. Wahlster, W. Conversational User Interfaces. Information Technology, Vol. 46 No. 6, (2004) 289 – 290.
3. Glass, J. et. al. A Framework for Developing Conversational User Interfaces. In Computer-Aided Design of User Interfaces IV, Springer, Netherlands. (2005) 349 – 360.
4. Zue, V & Glass, J. Conversational interfaces: Advances and challenges. Proceedings of the IEEE, Vol. 88, No.8, (2000) 1166 – 1180.
5. Kellner, A & Portele, T. SPICE: A Multimodal Conversational User Interface to an Electronic Program Guide, Philips Research Laboratories, Aachen (n.d.).
6. Guodong, Z. & Zhongjiao, J. (2010). From e-campus to e-learning: An overview of ICT applications in Chinese higher education. British Journal of Educational Technology, 41(4, (2010), 574 – 581.
7. Allen, K. Online learning: constructivism and conversation as an approach to learning. Innovations in Education and Teaching International, Vol. 42, No. 3, (2005) 247 – 256.
8. Boettcher, J. & Conrad, R. M. Faculty guide for moving teaching and learning to the web mission. Viejo, CA, League for Innovation in the Community College. (1999)
9. Chen, F. C. & Wang, T. C. Social conversation and effective discussion in online group learning. Education Tech Research Dev., Vol. 57, (2009) 587 – 612.
10. Rossett, A & Marshall, J. E-learning: what’s old is new again. T+D, January, (2010) 34 – 38.
11. Natural Language Processing, http://www.cnlp.org/publications/ 03nlp.lis.encyclopedia.pdf
12. Smith, D. Computerizing Computer Science. Communications of the ACM, 41, (1998) 21-23