Norma Ming

University of California, Berkeley, Education, Department Member

Carnegie Mellon University, Psychology, Graduate Student

Followers

Following

Public Views

Cognitive scientist, teacher, and researcher with K-16+ education experience spanning 20 years. Taught students from 6 years old through postgraduate level; trained teachers, school administrators, and faculty developers; developed and evaluated assessments, curricular materials, and technology-delivered instruction for NSF, ETS, and private education companies. Research projects encompass student learning in math and science, instructional strategies, educational technology, visual representations, collaborative learning, facilitating discussion, building professional communities.

Interests:
• Blending research with cutting-edge technology to create innovative, assessment-driven teaching and learning environments that combine machine intelligence with collaborative human intelligence to advance both insights and outcomes.
• Merging research, practice, and policy to improve how knowledge is developed, shared, and applied to yield better learning.
Supervisors: Ken Koedinger and Marsha Lovett

less

Related Authors

University of Bergen

University of Bahrain

Mihaela Cocea

University of Portsmouth

Steve Leichtweis

Alexander Wong

Joe Psotka

U.S. Army Research Laboratory

Duncan Ferguson

liaqat ali

Interests

Uploads

Papers by Norma Ming

Visualizing Topics, Time, and Grades in Online Class Discussions

We present a series of visualizations of online discussions that combine topic modeling with othe... more We present a series of visualizations of online discussions that combine topic modeling with other dimensions of the discussion contributions, to help faculty assess and improve learning from discussions. After applying probabilistic latent semantic analysis (pLSA) to calculate the relative conceptual distance between discussion posts, we projected posts or collections of posts into a two-dimensional space. By color-coding points according to their temporal position in the course or according to the author’s final grade, we captured patterns in students’ contributions that connect the topic modeling factors to more intuitively familiar characteristics. We consider how some possible qualitative features of the discussion may be represented in the topic space and outline future work to develop these tools further.

DownloadEdit

Automated Predictive Assessment from Unstructured Student Writing

Ming, N.C., & Ming, V.L. (2012a, September). Automated predictive assessment from unstructured student writing. Paper presented at the First International Conference on Data Analytics, Barcelona, Spain.

We investigated the validity of applying topic modeling to unstructured student writing from onli... more We investigated the validity of applying topic modeling to unstructured student writing from online class discussion forums to predict students’ final grades. Using only student discussion data from introductory courses in biology and economics, both probabilistic latent semantic analysis (pLSA) and hierarchical latent Dirichlet allocation (hLDA) produced significantly better than chance predictions which improved with additional data collected over the duration of the course. Hierarchical latent Dirichlet allocation yielded superior predictions, suggesting the feasibility of mining student data to derive conceptual hierarchies. Results indicate that topic modeling of student-generated text may offer useful formative assessment information about students’ conceptual knowledge.

DownloadEdit

Predicting student outcomes from unstructured data

Ming, N.C., & Ming, V.L. (2012). Predicting student outcomes from unstructured data. In J. Masthoff, B. Mobasher, M.C. Desmarais, & R. Nkambou (Eds.), Proceedings of 20th International Conference in User Modeling, Adaptation, and Personalization (UMAP). Montreal, Canada: Springer.

We investigated the validity of applying topic modeling to unstructured student text data from on... more We investigated the validity of applying topic modeling to unstructured student text data from online class discussion forums to predict students’ final grades. Using only student discussion data from introductory courses in biology and economics, both probabilistic latent semantic analysis (pLSA) and hierarchical la-tent Dirichlet allocation (hLDA) produced significantly better than chance predic-tions which improved with additional data collected over the duration of the course. Predictions were more accurate from hLDA than from pLSA, suggesting the feasibility and value of deriving conceptual hierarchies relevant to actual stu-dent data. Results indicate that topic modeling of student-generated text may pro-vide a useful source of formative assessment to support learning and instruction.

DownloadEdit

Using text mining to characterize online discussion facilitation

Facilitating class discussions effectively is a critical yet challenging component of instruction... more Facilitating class discussions effectively is a critical yet challenging component of instruction, particularly in online environments where student and faculty interaction is limited. Our goals in this research were to identify facilitation strategies that encourage productive discussion, and to explore text mining techniques that can help discover meaningful patterns in the discussions more efficiently at scale. Based on a close reading of selected discussion threads from online undergraduate science classes, we observed a variety of facilitation strategies associated with discussion quality. These observations informed our selection of a larger dataset of discussion threads to analyze via text mining techniques. Using latent semantic analysis to produce topic models of the content of the discussions, we constructed visualizations of the topical and temporal development of those discussions among students and faculty. These visualizations revealed patterns that appeared to correspond with specific facilitation styles and with the extent to which discussions remained focused on particular topics. From a case study focusing on six of these discussions, we documented distinct patterns in the types of facilitation strategies employed and the character of the discussions that followed. In our conclusion, we discuss potential applications of these analytical techniques for helping students, faculty, and faculty developers become more aware of their participation and influence in online discussions, thereby improving their value as a learning environment.

DownloadEdit

Analogies vs. contrasts: A comparison of their learning benefits

An ongoing challenge for educators is helping their students recognize connections between relate... more An ongoing challenge for educators is helping their students recognize connections between related information while also appreciating distinctions between only seemingly related information. Although instructing students to compare pairs of problems may help call attention to the appropriate features of the domain, the literature is mixed regarding the types of comparisons that are most beneficial, analogical comparisons or contrasting comparisons. Experiments on learning exploratory data analysis suggest a possible reconciliation of the competing claims, by distinguishing between the learning gains produced by each comparison type. By precisely assessing students’ knowledge before and after the comparisons, these studies reveal that analogical comparisons promote generalization across surface features, while contrasting comparisons promote discrimination between structures. Results indicate that these skills vary across individuals and should be assessed separately to optimally match instruction to their needs.

DownloadEdit

Data analysis skills: What and how are students learning?

DownloadEdit

Performance vs. learning: Knowing the right answers for the right reasons

An important dilemma to resolve in instruction is distinguishing between short-term performance a... more An important dilemma to resolve in instruction is distinguishing between short-term performance and long-term learning in assessing students’ progress. Conditions that appear favorable in acquisition are not always as effective at promoting subsequent retention and transfer, due to differences in the processing activities involved in training and at test; in some cases, poor performance in training produced better performance at test (Schmidt & Bjork, 1992). Since global measures of accuracy and speed are insufficient predictors of the effectiveness of training, we have used a modeling approach to draw inferences about the knowledge structures students use to solve problems both at test and in training. We will present the results from one study demonstrating the usefulness of such qualitative measures in predicting learning outcomes from training.

DownloadEdit

The impact of spurious correlations on students' problem-solving

Students are often susceptible to surface features when learning to solve problems in a new domai... more Students are often susceptible to surface features when learning to solve problems in a new domain. Providing example problems where salient surface features are spuriously correlated with the same problem type may encourage their use (Ben-Zeev & Star, 2001), whereas increasing the variability among superficial features during training may yield more robust knowledge (Schmidt & Bjork, 1992). To better understand the causes and consequences of this phenomenon, we compared the impact of two instructional regimens embodying these extremes and articulated detailed models of students’ surface and deep knowledge resulting from each training procedure, enabling us to distinguish between weak correct knowledge and strong incorrect knowledge.

DownloadEdit

Learning spurious correlations instead of deeper relations

Effective instructional design requires navigating the tradeoff between providing helpful cues to... more Effective instructional design requires navigating the tradeoff between providing helpful cues to the correct solutions and supplying hints that ultimately detract from what students learn. The present study manipulated the correlations between superficial features and the correct solutions in a set of training problems in the domain of exploratory data analysis and examined their effect on novices with no prior knowledge of statistics. Students who were trained on problems with these spurious correlations performed more poorly on posttest problems lacking these associations, making errors in the direction predicted by the misleading features. The theoretical and educational implications of the outcomes of this practice are discussed.

DownloadEdit

Perceptual learning modules in math & science instruction

Research in cognitive science suggests that perceptual learning-- changes in information extract... more Research in cognitive science suggests that perceptual learning-- changes in information extraction leading to efficient detection and classification of relevant pattern structure-- is a crucial component of expertise in many domains, including mathematics and science. Traditional learning formats do not do much to develop discovery or automaticity in the processing of structure, but appropriate conditions for doing so can be achieved using computer-based educational conditions for doing so can be achieved using computer-based educational technology, in the form of Perceptual Learning Modules (PLM's). Here we report progress in three domains using PLM's: university organic chemistry, multiple representations (graphs, equations, and word problems) of linear functions, and algebraic equivalences. These projects are in different stages of implementation, but objective data collected in two of them strongly support the value of PLM's in secondary school and university mathematics and science instruction.

DownloadEdit

Norma Ming

Uploads

Papers by Norma Ming

Log In