Abstract:

Nowadays, when education in all disciplines shifts from the traditional in-class format to blended-learning or flipped classrooms, tools for teaching scientific writing are becoming increasingly important. Writing research papers in English can be difficult even for native speakers, let alone, for foreigners as it requires both, good proficiency in English, and awareness of the international publishing standards.
Available software addressing the problem of scientific writing mostly includes tools that assess the writing quality related to lexical diversity, word frequency, and syntactic complexity, and guiding feedback to authors and/or provide for writing instructions, lexical items and academic phrases. Some of the tools offer automatic paraphrasing that actually could be dangerous for those who cannot judge the correctness of the output, which is far from being perfect.

There is a noticeable lack of tutoring systems that focus on automation of writing strategies and language support. We, therefore, have developed an intelligent educational tool, AutoTutor, which attempts just that. To achieve the benefits of one-on-one tutoring in a scalable and cost-effective manner we used interactive dialogue technologies with the use of NLP in STEM domains. All computer-human communication is done in the natural language.

The current AutoTutor realization supports speakers of Russian, but the methodology, algorithms and program modules can be used for tool implementations to support other non-English native speakers.

AutoTutor takes care of two main aspects of writing scientific papers:
a) ensuring the availability and correct structuring of all the “must-to-be-included” information to meet the international publishing standards.
b) ensuring correct English terminology and syntax, which are provided by a built-in Russian-English machine translation module and English NLG generation module, correspondingly.

In the course of a computer interview AutoTutor draws the author through all the necessary stages of writing a paper (title, abstract, related works, other sections, and conclusion). At each stage, the tool extracts from the author quantums of the paper content in the form of natural language phrases, while providing linguistic support in the form of English lexical menus, clichés, templates, Russian-English machine translation and automatic sentence generation.

Working with AutoTutor the author learns to fulfill the international publishing requirements, English terminology and syntax. The tool can be used in any format of education.

Keywords:

Scientific writing, intelligent tool, non-native English speakers, machine translation, natural language generation.