More Like this

1. Reader-scale Text Mining and Visualization in the Secondary Reader-scale Text Mining and Visualization in the Secondary Reader-scale Text Mining and Visualization in the Secondary ELA Classroom

   Nomura, Nichole; Korotkov, Georgii; Levine, Sarah; Bywater, Christine Wynn; Lee, Victor R (December 2024, The journal of interactive technology and pedagogy)

   We report on our work collaborating with two Bay-Area middle-school (grades 6, 7, and 8) districts to design, implement, and research the use of text visualizations in supporting English Language Arts and data literacy learning outcomes. Through co- design, classroom research, and participatory professional development with English Language Arts teachers, the project developed and tested text analytics tools, visualizations, and pedagogical routines. This article overviews the constraints we encountered, including technical constraints related to devices, access, and use, and pedagogical constraints including teacher and classroom time, and highlights a few strategies developed in response to those constraints. Drawing on classroom video, co- design sessions with teachers, and qualitative interviews, we find that “small” data visualizations that visualize text mining data created on single texts or small excerpts may support student data literacy and self-e!cacy as well as the development of a more expansive linguistic-experiential reservoir. We present a routine, “What did the computer miss?”, used by our collaborating teachers to support thinking with and about computing technologies that works especially well with single-/small- text data visualizations in the middle-school context, and conclude with suggestions and future directions for research and collaboration. 

   more » « less
2. Multi-scale mining of kinematic distributions with wavelets

   https://doi.org/10.21468/SciPostPhys.8.3.043  

   Lillard, Benjamin; Plehn, Tilman; Romero, Alexis; Tait, Tim (January 2020, SciPost Physics)

   Typical LHC analyses search for local features in kinematicdistributions. Assumptions about anomalous patterns limit them to arelatively narrow subset of possible signals. Wavelets extractinformation from an entire distribution and decompose it at all scales,simultaneously searching for features over a wide range of scales. Wepropose a systematic wavelet analysis and show how bumps, bump-dipcombinations, and oscillatory patterns are extracted. Our kinematicwavelet analysis kit KWAK provides a publicly available framework toanalyze and visualize general distributions. 

   more » « less
3. ASAP: Fast, Approximate Graph Pattern Mining at Scale

   Iyer, Anand Padmanabha; Liu, Zaoxing; Jin, Xin; Venkataraman, Shivaram; Braverman, Vladimir; Stoica, Ion (January 2018, USENIX Symposium on Operating Systems Design and Implementation (OSDI))
4. ASAP: Fast, Approximate Graph Pattern Mining at Scale

   Anand, I; Liu, Z; Jin, X; Venkataraman, S; Braverman, V; Stoic, I (January 2018, Proceedings of the USENIX conference)
5. Traditional kriging versus modern Gaussian processes for large‐scale mining data

   https://doi.org/10.1002/sam.11635  

   Christianson, Ryan B; Pollyea, Ryan M; Gramacy, Robert B (October 2023, Statistical Analysis and Data Mining: The ASA Data Science Journal)

   The canonical technique for nonlinear modeling of spatial/point‐referenced data is known as kriging in geostatistics, and as Gaussian Process (GP) regression for surrogate modeling and statistical learning. This article reviews many similarities shared between kriging and GPs, but also highlights some important differences. One is that GPs impose a process that can be used to automate kernel/variogram inference, thus removing the human from the loop. The GP framework also suggests a probabilistically valid means of scaling to handle a large corpus of training data, that is, an alternative to ordinary kriging. Finally, recent GP implementations are tailored to make the most of modern computing architectures, such as multi‐core workstations and multi‐node supercomputers. We argue that such distinctions are important even in classically geostatistical settings. To back that up, we present out‐of‐sample validation exercises using two, real, large‐scale borehole data sets acquired in the mining of gold and other minerals. We compare classic kriging with several variations of modern GPs and conclude that the latter is more economical (fewer human and compute resources), more accurate and offers better uncertainty quantification. We go on to show how the fully generative modeling apparatus provided by GPs can gracefully accommodate left‐censoring of small measurements, as commonly occurs in mining data and other borehole assays. 

   more » « less