mbsj 2011 poster バイオの買物.com

1
Innovations to Find the Best Products for Your Research Castle104 LLC., Naofumi Kagami website: www.castle104.com Twitter: @naofumi e-mail: [email protected] catalog.castle104.com, antibody.castle104.com "Search Engines should be replaced by Answer Engines" Steve Wozniak, co-founder Apple Computer 1. An ontology (categorization-scheme) to describe research tools. 2. A novel user-interface for easy browsing of thousands of ontologies (categories). 3. A semi-automated annotation system. 4. Product information retrieval system. What do we need to do? Designing a user interface suitable for browsing thousands of ontologies (categories) バイオの買物.com : まとめてカタログ Miller Columns まとめてカタログ implements Miller columns in combination with very fast search and multiple tags to drill down on facets. To our knowledge, this is the only implementation of dynamic Miller columns on a web interface. Semi-automated annotation Regular Expression synonym dictionary Keyword-based candidate collection メーカー横断検索だけでは不十分! Much more than just cross-maker search. Web scraping for product information retrieval 1. Find all antibodies for “CD100” which you can use on a flow cytometer FL2 channel. 2. Find all 100-bp ladders and sort them by price. 3. Find all real-time PCR reagents and discover new features. 4. List the various commercially available methods to analyze apoptosis. Usage Scenarios 1. Ruby, Ruby on Rails. 2. Solr (Sphinx) for fast database search. 3. Nested sets for hierarchical data management in SQL. 4. jQuery, Prototype.js for Miller column implementation. 5. Ruby Mechanize for web scraping. Key technologies 1. Cross-maker, annotated research tool database. 2. > 150,000 products, > 2,700 ontologies (categories). 3. Innovative user interface for browsing ontologies. 4. Roadmap: Increase maker participation and cover 90% of life-science reagents in 2012. 5. Our goal is to create a “PubMed” for research tools. Summary Search engines cannot answer these questions! Tree-navigation systems, despite being very popular for representing hierarchies, are often inadequate for complex ontologies. Nodes with many children can fill-up the display with irrelevant information, making the hierarchy almost impossible to understand. Tree Navigation AmiGO: Browse “PPAR pathway” Miller columns, although uncommon in web user interfaces, can handle complex hierarchies elegantly. In Miller columns, the hierarchy and the children are displayed independently on the horizontal and vertical axes respectively. In tree-navigation, both are mixed on the same vertical axis. Miller columns MacOS X browser: Rubygems folder for Research Tools news.castle104.com Seminar and campaign information from >100 supplier, university and research institute web sites. Anybody can upload information Manual curation of Seminar, Campaign information 1. Add date information. 2. Categorize. 3. Remove duplicates. Please contribute your information too! > 100 sources Automatic data retrieval 1. RSS feeds. 2. Web scraping. Campaigns listed by date Seminars listed by date (From both makers and researchers)

Upload: com-castle104-llc

Post on 06-Aug-2015

2.773 views

Category:

Technology


2 download

TRANSCRIPT

Page 1: MBSJ 2011 Poster バイオの買物.com

�������

Innovations to Find the Best Products for Your Research

Castle104 LLC., Naofumi Kagamiwebsite: www.castle104.comTwitter: @naofumie-mail: [email protected]

catalog.castle104.com, antibody.castle104.com

"Search Engines should be replaced by Answer Engines"

Steve Wozniak, co-founder Apple Computer

1. An ontology (categorization-scheme) to describe research tools.

2. A novel user-interface for easy browsing of thousands of ontologies (categories).

3. A semi-automated annotation system.4. Product information retrieval system.

What do we need to do?

Designing a user interface suitable for browsing thousands of ontologies (categories)

バイオの買物.com : まとめてカタログ Miller Columns

まとめてカタログ implements

Miller columns in combination with very fast search and multiple tags to drill down on facets.

To our knowledge, this is the only implementation of dynamic Miller columns on a web interface.

Semi-automated annotation

Regular Expression synonym dictionary Keyword-based candidate collection

メーカー横断検索だけでは不十分!Much more than just cross-maker search.

Web scraping for product information retrieval

1. Find all antibodies for “CD100” which you can use on a flow cytometer FL2 channel.

2. Find all 100-bp ladders and sort them by price.3. Find all real-time PCR reagents and discover new

features.4. List the various commercially available methods to

analyze apoptosis.

Usage Scenarios

1. Ruby, Ruby on Rails.2. Solr (Sphinx) for fast database search.3. Nested sets for hierarchical data management in SQL.4. jQuery, Prototype.js for Miller column implementation.5. Ruby Mechanize for web scraping.

Key technologies

1. Cross-maker, annotated research tool database.2. > 150,000 products, > 2,700 ontologies (categories).3. Innovative user interface for browsing ontologies.4. Roadmap: Increase maker participation and cover 90%

of life-science reagents in 2012.5. Our goal is to create a “PubMed” for research tools.

Summary

Search engines cannot answer these questions!

Tree-navigation systems, despite being very popular for representing hierarchies, are often inadequate for complex ontologies.Nodes with many children can fill-up the display with irrelevant information, making the hierarchy almost impossible to understand.

Tree Navigation

AmiGO: Browse “PPAR pathway”

Miller columns, although uncommon in web user interfaces, can handle complex hierarchies elegantly.In Miller columns, the hierarchy and the children are displayed independently on the horizontal and vertical axes respectively. In tree-navigation, both are mixed on the same vertical axis.

Miller columns

MacOS X browser: Rubygems folder

for Research Tools

news.castle104.com

Seminar and campaign information from >100 supplier, university and research institute web sites.

Anybody can upload information

Manual curation of Seminar, Campaign information1. Add date information.2. Categorize.3. Remove duplicates.

Please contribute your information too!

> 100 sources

Automatic data retrieval1. RSS feeds.2. Web scraping.

Campaigns listed by date Seminars listed by date (From both makers and researchers)