A Comparative Study on Various Vocabulary Knowledge Scales for Predicting Vocabulary Pre-Knowledge

1. Introduction

We have encountered and witnessed the great popularity of various emerging e-learning resources such as massive open online courses (MOOCs), textbooks and videos with the development of the big data era. On the other hand, the emerging social media and ubiquitous techniques have played an important role in our daily lives. These evolving techniques and e-learning applications have not only provided a large volume of learning resources on the web but also significantly changed the ways people learn new knowledge and access information.

It is critical to understand the characteristics of users to assist them in finding desired and relevant learning resources in such a large volume of resources. For example, understanding the pre-knowledge on vocabulary of learners is very prominent and useful for language learning systems (Chen & Li, 2010; Zou, Xie, Li, Wang & Chen, 2014). That is, the language learning effectiveness can be significantly improved if the pre-knowledge levels of learners on vocabulary can be accurately predicted. Another typical example of modeling the characteristics of users is the tag-based user profiles which have widely adopted to facilitate personalized search or recommendation in the social media domain (Xie, Li & Mao 2012; Xie, Li, Mao, Li, Cai & Rao 2014).

In this research, we focus on the research problem of how to model language learners in the e-learning systems. As the fast development and popularity of e-learning systems, a collection of user learning documents such as writing assignments, reading essays and accessed lecture notes are available in log database of such systems (Li, Lau, Wah, Ashman, Leung, Li & Lee, 2009). To understand the language learners, we try to predict the pre-knowledge levels on vocabulary of learners according to their history of learning documents. In particular, we focus on two kinds of learning documents, which are writing assignments and reading essays. The reason of selecting writing assignments and reading essays is that they can reflect both receptive and productive knowledge of words (Webb, 2005). The main contributions of this research are summarized as follows.

• 1.

  We propose three rating functions according to different granularity of the vocabulary knowledge scales (VKS) for identifying pre-knowledge levels on vocabulary of learners;

• 2.

  We model the language learner by adopting vector-based learner profile;

• 3.

  We present a framework which can integrate with different rating functions for writing assignments and reading essays;

• 4.

  We compare accuracy of the proposed framework with baselines in real-world study circumstances;

• 5.

  We examine the impact of various parameters and document sources in the experiment.

The rest of this article is organized as follows. In Section 2, we review the related research on language learning and user modeling. In Section 3, we introduce the proposed rating functions, framework and models. Section 4 reports the experimental settings, metrics and results. The implications of this research are discussed in Section 5. In Section 6, the findings of this research are summarized and future directions are discussed.

2. Related Work

In this section, we review the related research in two areas, which are user modeling techniques, and language learning systems.