The source code for version 2.0 of the khmer project. http://khmer.readthedocs.org/
Folk taxonomies are powerful cultural tools for the categorization and utilization of the world i... more Folk taxonomies are powerful cultural tools for the categorization and utilization of the world in which a people live. The English language, for example, has a few folk taxa remaining; including pets, farm animals, and evergreens. Folk taxa are categories or logical groupings, usually referring to nature, which may have social and cultural relevance, but not necessarily possessing any scientific relatedness amongst their members. They are useful in day-to-day dealings with the environment, providing a catalogue grouped by salient features. Finding a language's folk taxonomy can often be difficult, with the lines drawn between categories often not readily apparent. With this work I examine the theory behind folk taxonomic classification and attempt to devise methods for unearthing folk taxonomies with the help of Natural Language Processing. The subject language of this inquiry is Matukar. Matukar is an Austroneasian language of Papua New Guinea, spoken by only about 430 village...
The underlying assumption behind Hadoop and, more generally, the need for distributed processing ... more The underlying assumption behind Hadoop and, more generally, the need for distributed processing is that the data to be analyzed cannot be held in memory on a single machine. Today, this assumption needs to be re-evaluated. Although petabyte-scale datastores are increasingly common, it is unclear whether “typical” analytics tasks require more than a single high-end server. Additionally, we are seeing increased sophistication in analytics, e.g., machine learning, where we process smaller and more refined datasets. To address these trends, we propose “scaling down” Hadoop to run on multi-core, shared-memory machines. This paper presents a prototype runtime called Hone (“Hadoop One”) that is API compatible with Hadoop. With Hone, we can take an existing Hadoop application and run it efficiently on a single server. This allows us to take existing MapReduce algorithms and find the most suitable runtime environment for execution on datasets of varying sizes. For dataset sizes that fit int...
Folk taxonomies are powerful cultural tools for the categorization and utilization of the world i... more Folk taxonomies are powerful cultural tools for the categorization and utilization of the world in which a people live. The English language, for example, has a few folk taxa remaining; including 'fruits', 'vegetables', 'pets', 'farm animals', and 'evergreens'. Folk taxa are categories or logical groupings, usually referring to nature, which may have social and cultural relevance, but not necessarily possessing any scientific relatedness amongst their members. They are useful in day-today dealings with the environment, providing a catalogue grouped by salient features. With the assistance of Natural Language Processing the examination of folk taxonomies may be streamlined, providing linguists with a starting point with which to theorize folk taxa. I apply these tools to the Matukar Language, and examine the results.
The underlying assumption behind Hadoop and, more generally, the need for distributed processing ... more The underlying assumption behind Hadoop and, more generally, the need for distributed processing is that the data to be analyzed cannot be held in memory on a single machine. Today, this assumption needs to be re-evaluated. Although petabyte-scale data-stores are increasingly common, it is unclear whether "typical" analytics tasks require more than a single high-end server. Additionally, we are seeing increased sophistication in analytics, e.g., machine learning, which generally operates over smaller and more refined datasets. To address these trends, we propose "scaling down" Hadoop to run on shared-memory machines. This paper presents a prototype runtime called Hone, intended to be both API and binary compatible with standard (distributed) Hadoop. That is, Hone can take an existing Hadoop jar and efficiently execute it, without modification, on a multi-core shared memory machine. This allows us to take existing Hadoop algorithms and find the most suitable run-t...
Page 1. 1 NLP Analysis of Folksonomies An examination of the Matukar language Jonathan Gluck 1 Ab... more Page 1. 1 NLP Analysis of Folksonomies An examination of the Matukar language Jonathan Gluck 1 Abstract Folk taxonomies are powerful cultural tools for the categorization and utilization of the world in which a people live. The English language, for example, ...
Sentiment analysis, a powerful tool determining public opinion, is often applied to corpora of fu... more Sentiment analysis, a powerful tool determining public opinion, is often applied to corpora of full length documents. Guided by recent forays into sentiment analysis on micro-blogging platforms, we will examine popular opinion of movies and stocks and compare them to real world indicators of popularity(e.g: box office.
Page 1. 1 NLP Assisted Analysis of Folk Taxonomy An examination of the Matukar language Jonathan ... more Page 1. 1 NLP Assisted Analysis of Folk Taxonomy An examination of the Matukar language Jonathan Gluck Abstract Folk taxonomies are powerful cultural tools for the categorization and utilization of the world in which a people live. The English language, for example, ...
Energy efficient computing is growing in importance as a means to cut costs and conserve power. O... more Energy efficient computing is growing in importance as a means to cut costs and conserve power. One recent approach to reducing the energy cost of data management is to use large clusters of energy efficient but slow computers rather than smaller clusters of more conventional high-power machines. For these systems to perform similarly to high-power machines but on slower computers they often support only the simple data interface of a key-value storage system, which can incur high communication costs if sophisticated data operations are needed. In this project we extend one existing lowpower system, FAWN-KV, to support more sophisticated sever-side data operations. We added support for lists, sets, and maps and compared the performance of our modifications to the performance of unmodified FAWN-KV on data operations like those used in modern systems. Our modifications improved performance by up to a factor of three and outperformed the basic system in all tests.
Business intelligence”, a term first coined in 1958 [11], has been evolving for almost 60 years. ... more Business intelligence”, a term first coined in 1958 [11], has been evolving for almost 60 years. With recent advancements in data collection [15, 2, 8, 13], companies are looking for new ways to gain insight into customer behavior. In this paper, we present C-Flow, a novel system that allows mall and store owners to reason about consumer foot traffic with respect to store profits. This is done with an integrated heat map and overlaid footpaths on a mall map. Through two rounds of usability studies, we also provide evidence that rendering abstract paths may be more useful than rendering raw path data. See video at http://goo.gl/mBXjKd for reviews, slides and demonstration. Author
The khmer package is a freely available software library for working efficiently with fixed lengt... more The khmer package is a freely available software library for working efficiently with fixed length DNA words, or k-mers. khmer provides implementations of a probabilistic k-mer counting data structure, a compressible De Bruijn graph representation, De Bruijn graph partitioning, and digital normalization. khmer is implemented in C++ and Python, and is freely available under the BSD license at https://github.com/dib-lab/khmer/.
The source code for version 2.0 of the khmer project. http://khmer.readthedocs.org/
Folk taxonomies are powerful cultural tools for the categorization and utilization of the world i... more Folk taxonomies are powerful cultural tools for the categorization and utilization of the world in which a people live. The English language, for example, has a few folk taxa remaining; including pets, farm animals, and evergreens. Folk taxa are categories or logical groupings, usually referring to nature, which may have social and cultural relevance, but not necessarily possessing any scientific relatedness amongst their members. They are useful in day-to-day dealings with the environment, providing a catalogue grouped by salient features. Finding a language's folk taxonomy can often be difficult, with the lines drawn between categories often not readily apparent. With this work I examine the theory behind folk taxonomic classification and attempt to devise methods for unearthing folk taxonomies with the help of Natural Language Processing. The subject language of this inquiry is Matukar. Matukar is an Austroneasian language of Papua New Guinea, spoken by only about 430 village...
The underlying assumption behind Hadoop and, more generally, the need for distributed processing ... more The underlying assumption behind Hadoop and, more generally, the need for distributed processing is that the data to be analyzed cannot be held in memory on a single machine. Today, this assumption needs to be re-evaluated. Although petabyte-scale datastores are increasingly common, it is unclear whether “typical” analytics tasks require more than a single high-end server. Additionally, we are seeing increased sophistication in analytics, e.g., machine learning, where we process smaller and more refined datasets. To address these trends, we propose “scaling down” Hadoop to run on multi-core, shared-memory machines. This paper presents a prototype runtime called Hone (“Hadoop One”) that is API compatible with Hadoop. With Hone, we can take an existing Hadoop application and run it efficiently on a single server. This allows us to take existing MapReduce algorithms and find the most suitable runtime environment for execution on datasets of varying sizes. For dataset sizes that fit int...
Folk taxonomies are powerful cultural tools for the categorization and utilization of the world i... more Folk taxonomies are powerful cultural tools for the categorization and utilization of the world in which a people live. The English language, for example, has a few folk taxa remaining; including 'fruits', 'vegetables', 'pets', 'farm animals', and 'evergreens'. Folk taxa are categories or logical groupings, usually referring to nature, which may have social and cultural relevance, but not necessarily possessing any scientific relatedness amongst their members. They are useful in day-today dealings with the environment, providing a catalogue grouped by salient features. With the assistance of Natural Language Processing the examination of folk taxonomies may be streamlined, providing linguists with a starting point with which to theorize folk taxa. I apply these tools to the Matukar Language, and examine the results.
The underlying assumption behind Hadoop and, more generally, the need for distributed processing ... more The underlying assumption behind Hadoop and, more generally, the need for distributed processing is that the data to be analyzed cannot be held in memory on a single machine. Today, this assumption needs to be re-evaluated. Although petabyte-scale data-stores are increasingly common, it is unclear whether "typical" analytics tasks require more than a single high-end server. Additionally, we are seeing increased sophistication in analytics, e.g., machine learning, which generally operates over smaller and more refined datasets. To address these trends, we propose "scaling down" Hadoop to run on shared-memory machines. This paper presents a prototype runtime called Hone, intended to be both API and binary compatible with standard (distributed) Hadoop. That is, Hone can take an existing Hadoop jar and efficiently execute it, without modification, on a multi-core shared memory machine. This allows us to take existing Hadoop algorithms and find the most suitable run-t...
Page 1. 1 NLP Analysis of Folksonomies An examination of the Matukar language Jonathan Gluck 1 Ab... more Page 1. 1 NLP Analysis of Folksonomies An examination of the Matukar language Jonathan Gluck 1 Abstract Folk taxonomies are powerful cultural tools for the categorization and utilization of the world in which a people live. The English language, for example, ...
Sentiment analysis, a powerful tool determining public opinion, is often applied to corpora of fu... more Sentiment analysis, a powerful tool determining public opinion, is often applied to corpora of full length documents. Guided by recent forays into sentiment analysis on micro-blogging platforms, we will examine popular opinion of movies and stocks and compare them to real world indicators of popularity(e.g: box office.
Page 1. 1 NLP Assisted Analysis of Folk Taxonomy An examination of the Matukar language Jonathan ... more Page 1. 1 NLP Assisted Analysis of Folk Taxonomy An examination of the Matukar language Jonathan Gluck Abstract Folk taxonomies are powerful cultural tools for the categorization and utilization of the world in which a people live. The English language, for example, ...
Energy efficient computing is growing in importance as a means to cut costs and conserve power. O... more Energy efficient computing is growing in importance as a means to cut costs and conserve power. One recent approach to reducing the energy cost of data management is to use large clusters of energy efficient but slow computers rather than smaller clusters of more conventional high-power machines. For these systems to perform similarly to high-power machines but on slower computers they often support only the simple data interface of a key-value storage system, which can incur high communication costs if sophisticated data operations are needed. In this project we extend one existing lowpower system, FAWN-KV, to support more sophisticated sever-side data operations. We added support for lists, sets, and maps and compared the performance of our modifications to the performance of unmodified FAWN-KV on data operations like those used in modern systems. Our modifications improved performance by up to a factor of three and outperformed the basic system in all tests.
Business intelligence”, a term first coined in 1958 [11], has been evolving for almost 60 years. ... more Business intelligence”, a term first coined in 1958 [11], has been evolving for almost 60 years. With recent advancements in data collection [15, 2, 8, 13], companies are looking for new ways to gain insight into customer behavior. In this paper, we present C-Flow, a novel system that allows mall and store owners to reason about consumer foot traffic with respect to store profits. This is done with an integrated heat map and overlaid footpaths on a mall map. Through two rounds of usability studies, we also provide evidence that rendering abstract paths may be more useful than rendering raw path data. See video at http://goo.gl/mBXjKd for reviews, slides and demonstration. Author
The khmer package is a freely available software library for working efficiently with fixed lengt... more The khmer package is a freely available software library for working efficiently with fixed length DNA words, or k-mers. khmer provides implementations of a probabilistic k-mer counting data structure, a compressible De Bruijn graph representation, De Bruijn graph partitioning, and digital normalization. khmer is implemented in C++ and Python, and is freely available under the BSD license at https://github.com/dib-lab/khmer/.
Uploads
Papers by Jonathan Gluck