Research Of Applications Of Cognitive Computing In The Area Of Education

Cognitive computing is the new area of Artificial Intelligence (AI), relying on techniques based on expert systems and also statistics and mathematical models. In short, cognitive computing systems can be regarded as a “more human” artificial intelligence. In fact, they emulate human thinking procedures, showing special capabilities in dealing with uncertainties and in solving problems that typically entail computation consuming processes. They can evolve, exploiting the accumulated experience to learn from the past, both from errors and from successful findings. From a theoretical point of view, cognitive computing can replace today’s calculators in many fields of application but therefore hardware requirements are very high. In this paper, we focus on benefits that cognitive computing technology can bring when applied in the education field and we make a short review of relevant experiences.

Introduction

Getting something new which is offered by cognitive computing, we can imagine the raise of a new generation of automated Information Technology (IT) systems committed to relief humans from solving a variety of problems. They will be based on human cognitive capabilities simulation, through machine learning algorithms inherited from the field of Artificial Intelligence (AI). In detail, we can identify three main areas in which cognitive computing has significant role, that are:

• Advancements in computation capabilities
• Human-computer interactions and communication
• The evolution of the Internet of Things (IoT).

Cognitive computing can successfully deal with complex tasks such as, e. g. , knowledge management, natural language processing and classification, data mining, hence cognitive systems can successfully perform duties such as relationships extraction from unstructured corpus, speech-to-text and text-to-speech conversions, pattern recognition, computer vision, machine learning, which involve research areas based on human-like reasoning techniques. From a handy perspective, cognitive computing systems owe their capacity to the extensive measure of information they analyze and process to feed the machine learning algorithms they are depended on. As another example, let us consider the recommendation mechanisms adopted by some electronic commerce sites, which apply sophisticated machine learning algorithms to provide users with suggestions about other products that are related to the ones they are viewing or have put in their wish lists, based on their preferences and on that item’s purchase history. Cognitive computing cultivates better approaches for association among people and PCs. Cognitive computing simulates human reasoning processes translating them in suited templates, it can help machines to learn and to teach humans new concepts and behaviors. To conclude, given these premises, we argue that cognitive computing will be one of the main drivers for the near future developments in many fields of application. In fact, there are many possibility waves to surf, which can drive the evolution in the profession of IT.

According to this general scenario, we are going to investigate how cognitive computing with related technologies and applications impact on education.

Specifically, according to the smarter university model, cognitive computing-based applications and services should be taken for administration and management, and learning activities as well. From the e-learning point of view, we report some experiences that which are evaluated how cognitive computing can be an accelerator for students’ achievements, and a valuable support for the teachers. In particular:

• integrating cognitive computing services in software applications can strongly enhance students’ performances in computer science classes;
• studying cognitive computing behavior can lead to significant results in AI related studies;
• using a cognitive computing layer for digital interactions with students can enhance their performances and ease the teachers’ job in managing classes and learning materials.

Cognitive computing services in software development

We report an interesting experience of using cognitive computing in educational systems, that is the Watson Student Showcase organized by IEEE jointly with IBM. Students were challenged in a competition where they had to develop apps using the cognitive computing services included in Bluemix. The first five classified cognos (the cognitive apps) were attributed a $ 2, 000 award each and the assessment was based on specific features the apps were required to expose, such as: originality, feasibility, usefulness, and creativity. Students involved in the competition delivered the following applications:

• docbot, a mobile app that summarizes Electronic Health Record (EHR) information to streamline medical appointments. In this app, cognitive computing capabilities are used to extract and organize the most relevant information for the situation, so to provide physicians only with data relevant the required medical examination. Moreover, the application interface is mainly based on natural language processing;
• wordinator, which aims to facilitating the search for the most suitable word to best express own thoughts. This is also an implementation of the available natural language processing services;
• miface, eminently capable of capturing facial expressions, those who assume in particular situations (anger, fear, surprise, fear, loss, etc. ) although numerous expressions are already present in many algorithms, some of these need to be further analyzed by the human being to not generate false positives;
• telephony, which is based on the classical children game called “telephone”. In this game a group of friends in turn whisper something in the ear of his neighbor and finally the last one has to infer what was whispered by the first player. Except that in this application the circle of friends does not really exist and the message is repeatedly transformed by the cognitive system engine;
• stack analyze, which was designed to relief moderators of online communities from the heavy task of reviewing hundreds and hundreds of messages to determine whether a question of a community member is appropriate or not.

Another experience carried on in Italy demonstrated how the use of the IBM Watson services within the IBM Bluemix platform can enhance students’ performances in computer programming classes. more detail, the research investigates how students’ achievements varied depending on the adopted technological infrastructure, in developing similar projects, based on fundamentals of computer programming, data structures and algorithm. To this aim, past collaborative learning experiences, which were used as the starting point and reference to experiment teaching computer programming in a Platform as a Service (PaaS) environment. Among the many services made available within the platform, most students have profitably used those based on cognitive computing, to empower the functionalities of their demo projects, including, e. g. , speech recognition, automatic translation, file conversion, text-to-speech services for accessibility matters, and many others.

In conclusion, in this context, we were able to observe that students gained core competences in computer programming faster and improved the overall quality of their products owing to better interactions and natural interfaces. The above-cited cases demonstrate with practical learning tasks that cognitive computing can be a paramount support in computer science laboratories and related activities.

Studying cognitive systems behavior

Another interesting application of cognitive computing in education is reported, which analyze and try to understand some of the Watson’s functionalities. Specifically, they describe a text analytics course focused on building a simulator of IBM Watson, conducted in Spring 2014 at UNC Charlotte. They state that it is the first time that such a simulation was created in a university classroom. Their research also reports that the simulated system almost achieved a 20% accuracy on the Jeopardy! questions, which were used as a benchmark.

Cognitive computing supporting teachers

Another aspect is related to the support that teachers could get from cognitive computing-based applications in solving common students’ issues such as, e. g. , school dropout, individualization of learning, customization of training path, etc. , due to their capability of analyzing data. According to a novel paradigm, the teacher can ask the cognitive computing-based system to “talk” with students, in order to understand what are individual strengths and weaknesses and possible issues to face up to. These will be analyzed and reported to the teacher who then will decide the best strategies, course materials, etc. , to overcome those issues. This feature is natively implemented in Watson and it is known as Ask Watson. Specifically, the cited example represents a triage student-Watson-teacher, which makes applicable the individual learning model.

In this respect, one can ask to discover a student’s style of learning e. g. , visual, auditory. Otherwise, one can ask to prepare a small subclass for the math or computer science Olympics so that only students with specific attitudes will be selected and then will be provided with suited material and proper training to achieve their objectives at the best of their possibilities. In more detail, the above-cited paradigm consists of recursive cycles throughout 6 steps:

1. the cognitive system identifies weaknesses and strengths of every student;
2. the cognitive system recommends behavior and contents for students aligned to their skills and learning styles;
3. the teacher selects an appropriate learning path and creates an individualized plan for each student;
4. the students use the recommended content from the plan material;
5. the teacher monitors students’ progresses and, possibly, makes slight adjustments to the plan;
6. The teacher uses cognitive computing system to identify students’ skills attainment aligned to the defined standards.

Moreover, it is worthwhile noticing that, beside advanced reasoning and decision making capabilities, the interaction mechanism based on spoken language assumes a paramount importance. Specifically, the interaction consists of three steps, mapped on the relevant cognitive services:

1. 1ask,
2. discover,
3. decide, delivered in sequence.

The steps are

1. The student asks the cognitive system to be guided, through the speech-to-text and text-to-speech services;
2. The cognitive system teaches the student how to derive the answers from the questions asked through the natural language classifier and relationship extraction services;
3. The teacher sees the results through the tradeoff analytics and delegates the cognitive system to adapt the curriculum.

This mode of assistance is very promising and is already providing fruits. The application of the cognitive system has been used in (i) Wichita State University, which uses advanced analytics to predict potential students’ chances of success: 15% boost in registration. (ii) Hamilton County, Tennessee’s Department of Education, which uses predictive analytics to improve student achievement causing the following results: 8% increase in the graduation rate to 80% and 25% reduction in the annual drop rate; Finally, we also cite the experience carried on by Sundararajan and Nitta (2015), which designed and realized a tutoring system for K12 students, intensively using interactivity, automatic generation of questions, and learning analytics.

Conclusions

As a conclusion, after this short survey of cognitive computing technologies applied to education, our findings are encouraging. In fact, from this preliminary analysis emerges that there is a plenty of possibilities for the use of novel cognitive computing-based solutions and cognos for improving current performances of software, as well as to design novel applications, based on a new approach and on a new human-machine interaction and communication paradigm. However, this scenario seems to be far to realize, due to the lack of infrastructural settings and to the availability of open big data in general. Moreover, we notice that the reported experiences are from United States where the education system, and related business, is quite different from the majority of other countries. As an example, let us consider the Italian situation. At the best of the authors’ knowledge, cognitive computing and related issues was part of computer programming classes in a few Italian universities, i. e. , Genoa, Florence, and two in Naples that is “Federico II” and “Seconda Università”. They presented their findings and interesting students’ project within the workshop Head in the Clouds, focused on cloud computing and PaaS, i. e. , Eclipse meets Bluemix, held within the X annual Workshop of the Italian Eclipse Community. In addition, as a drawback, we found that one of the main problems is how to effectively train a cognitive system, which is very important for its responses accuracy. In this respect, Murtaza et al. (2016) showed that if the cognitive system is trained with well segmented documents, with semantically relevant titles and content consistent with the titles, then the accuracy of the response may even exceed 95%.

Given these premises, we envisage a rapid development of cognitive computing technology and related applications in the near future but this will be possible only in a complex ecosystem where the valuable data that are available within the education system, including institutions and municipalities, can be exchanged between applications and between organizations in a seamless and transparent manner.