IAnnSolrMigration

From Protein Prediction 2 Winter Semester 2014
Revision as of 21:46, 1 January 2015 by Ppwikiuser (talk | contribs) (Solutions required)


Introduction


iAnn is a platform which collects and distributes the data about various events, like symposiums, presentations, workshops, conferences etc.

The aim of iAnn platform is:

  • To save time, effort and cost of manual announcement curation for scientific organizations.
  • To facilitate dissemination, sharing and promotion of scientific announcements.
  • To avoid redundancy in manual curation of announcement data.

Problem description

While fetching the data from different sources, one needs to standardize it and name all properties in such way that name itself is descriptive and makes sense. Naming examples: submission_date, keywords, host_institution.

However, naming is part of agreement from the SASI document [sasi link] and it is straight forward. The challenge is to migrate all imported data from old to new schema. Solr does not provide automatized way and custom scripts have to be created. While working on customized scripts we considered several points;

  • scripts must be easy to modify
  • they must handle different types conversions while migrating
  • they must be able to handle different naming structures
  • they must provide a sort of feedback while migrating (process)
  • scripts must be configured on linode virtual server
  • migration from solr 4.1.0 to 4.10.2 should be done

Solutions required

We decided to divide our work to two separate tasks:

  • Creating/configuring SASI solr schema and upgrading solar service to new version
  • Writing scripts for data reindexing
  • Creating Web tool for easier control during migration
Project 1: Creating/configuring SASI solr schema and upgrading solar service to new version

We considered SASI document (sasi03.xml) as agreement for the naming structure of our data API. Old schema has to be updated and new collection in database has to be created. On top of that, solr service has to be reconfigured and migrated from older version 4.1.0 to 4.10.2.

Project 2: Writing scripts for data reindexing

After having the new schema and upgraded solr running, old data has to be fetched, modified and saved to the new collection. For this purpose we focused on creating scripts for solr migration.

Project 3: Creating Web tool for easier control during migration

To make everything more stable we focused on building a Web user interface where user can trigger migration in a browser. A tool should have RESTful architecture and it should be able to communicate with web server and guide user through the process. It should also allow user to connect old and new names from schema and auto-type detection and conversion. Example of automated data detection and conversion would be the case when we want to migrate array field keywords to the string. In this case we should detect array type, convert it to the string and then copy to new collection.

Project progress


Project progress
Week number Completed tasks Status
Week 1
  • Collecting ideas for possible visualization components that can be incorporated in iAnn viewer.
  • Understanding the working of iAnn viewer.
  • Underwent a tutorial on D3 javascript.
  • Vreatinga GitHub Repository.
Check.jpg
Week 2
  • Finalizing the ideas.
  • Developing the word cloud based on Host, Keyword, Provider and Country.
  • Integrating the word cloud with iAnn Viewer.
  • Initial Project setup for 3D Globe.
Check.jpg
Week 3
  • Enhancing the word cloud.
  • Bringing 3D Globe into a working condition.
  • Initial project setup for Partition Layout .
Check.jpg
Week 4
  • Word Cloud Completion.
  • Adding tooltip and list of events to each country.
  • Adding Markers for 3D Globe.
Check.jpg
Week 5
  • Incorporating 2 options : Ortho(3D) and Equirectangular(2D) for better user interface.
  • Automatic Zoom in.
  • Stabilizing the 3D to 2D transition.
  • Completion of 3D Globe.
Check.jpg
Week 6

Proposal for Schema changes


Current version of iAnn website is delivering a service by using custom schema for data attributes naming. However, it is not standardized and does not have default field naming scheme. Therefore, a community agreed on default schema layout, named SASI. In this document we will point out differences and describe new schema fields. As current iAnn collection runs on Apache Solr 4.1 and the latest one with some major bug fixes is 4.10.2, we decided to migrate and update our service as well.

Accomplished tasks:

  • Understanding of existing scheme,
  • Proposing structure changes and
  • Highlighting differences


Old Scheme:

  <field name="id" type="string" indexed="true" stored="true" required="true" />
  <field name="title" type="text" indexed="true" stored="true"/>
  <field name="subtitle" type="text" indexed="true" stored="true"/>
  <field name="description" type="string" indexed="false" stored="true"/>
  <field name="provider" type="text_lowercase" indexed="true" stored="true"/>
  <field name="link" type="string" indexed="false" stored="true"/>
  <field name="start" type="tdate" indexed="true" stored="true"/>
  <field name="end" type="tdate" indexed="true" stored="true"/>
  <field name="venue" type="text_lowercase" indexed="true" stored="true"/> 
  <field name="city" type="text_lowercase" indexed="true" stored="true"/>
  <field name="county" type="string" indexed="fasle" stored="true"/>
  <field name="country" type="text_lowercase" indexed="true" stored="true"/>
  <field name="postcode" type="string" indexed="false" stored="true"/>
  <field name="attachment" type="string" indexed="false" stored="true" multiValued="true"/>
  <field name="image" type="string" indexed="false" stored="true" multiValued="true"/>
  <field name="keyword" type="text_lowercase" indexed="true" stored="true" multiValued="true"/>
  <field name="category" type="text_lowercase" indexed="true" stored="true" multiValued="true"/>
  <field name="field" type="text_lowercase" indexed="true" stored="true" multiValued="true"/> 
  <field name="submission_name" type="text_lowercase" indexed="true" stored="true" multiValued="true"/>
  <field name="submission_email" type="string" indexed="true" stored="true" multiValued="true"/>
  <field name="submission_date" type="tdate" indexed="true" stored="true" multiValued="true"/>
  <field name="submission_comment" type="text" indexed="false" stored="true" multiValued="true"/>
  <field name="submission_organization" type="text" indexed="true" stored="true" multiValued="true"/>
  <field name="latitude" type="double" indexed="true" stored="true"/>
  <field name="longitude" type="double" indexed="true" stored="true"/>
  <field name="sponsor" type="text_lowercase" indexed="true" stored="true"/>
  <field name="public" type="boolean" indexed="true" stored="true"/>
  <field name="text" type="text" indexed="true" stored="false" multiValued="true"/>
  <field name="_version_" type="long" indexed="true" stored="true"/>


New Scheme:

  <field name="id" type="text" indexed="true" stored="true" required="true" />
  <field name="title" type="text" indexed="true" stored="true"/>
  <field name="subtitle" type="text" indexed="true" stored="true"/>
  <field name="description" type="text" indexed="false" stored="true"/>
  <field name="prerequisites" type="text" indexed="false" stored="true"/>
  <field name="programme" type="programme" indexed="false" stored="true"/>
  <field name="comments" type="text" indexed="false" stored="true"/>
  <field name="fees" type="fees" indexed="false" stored="true" multiValued="true"/>
  <field name="discount" type="discount" indexed="false" stored="true"/>
  <field name="accreditation" type="text" indexed="false" stored="true"/>
  <field name="status" type="text" indexed="false" stored="true"/>
  <field name="eligibility" type="text" indexed="false" stored="true" multiValued="true"/>
  <field name="capacity" type="int" indexed="false" stored="true"/>
  <field name="contact" type="person" indexed="false" stored="true" multiValued="true"/>
  <field name="submitter" type="person" indexed="false" stored="true" multiValued="true"/>
  <field name="organizers" type="person" indexed="false" stored="true" multiValued="true"/>
  <field name="speakers" type="person" indexed="false" stored="true" multiValued="true"/>
  <field name="host_institution" type="organization" indexed="false" stored="true" multiValued="true"/>
  <field name="sponsor" type="organization" indexed="false" stored="true" multiValued="true"/>
  <field name="venue" type="text" indexed="false" stored="true"/>
  <field name="street_address" type="text" indexed="false" stored="true"/>
  <field name="city" type="text" indexed="false" stored="true"/>
  <field name="province" type="text" indexed="false" stored="true"/>
  <field name="country" type="text" indexed="false" stored="true"/>
  <field name="postcode" type="text" indexed="false" stored="true"/>
  <field name="post_office_box" type="text" indexed="false" stored="true"/>
  <field name="url" type="link" indexed="false" stored="true"/>
  <field name="attachment" type="link" indexed="false" stored="true" multiValued="true"/>
  <field name="social_media" type="link" indexed="false" stored="true" multiValued="true"/>
  <field name="starts" type="date" indexed="false" stored="true"/>
  <field name="ends" type="date" indexed="false" stored="true"/>
  <field name="time_zone" type="text" indexed="false" stored="true"/>
  <field name="last_update" type="date" indexed="false" stored="true" multiValued="true"/>
  <field name="deadlines" type="date" indexed="false" stored="true" multiValued="true"/>
  <field name="registration_opens_date" type="date" indexed="false" stored="true" multiValued="true"/>
  <field name="acceptance_notification_date" type="date" indexed="false" stored="true" multiValued="true"/>
  <field name="type" type="text" indexed="false" stored="true" multiValued="true"/>
  <field name="topic" type="text" indexed="false" stored="true" multiValued="true"/>
  <field name="public" type="boolean" indexed="false" stored="true"/>
  <field name="target_audience" type="text" indexed="false" stored="true" multiValued="true"/>
  <field name="spotlight" type="boolean" indexed="false" stored="true"/>
  <field name="latitude" type="double" indexed="true" stored="true"/>
  <field name="longitude" type="double" indexed="true" stored="true"/>
  <field name="_version_" type="long" indexed="true" stored="true"/>


Conclusion:

While using newly proposed scheme iAnn will have a possibility to serve even more diverse data. This will serve as a basis for new services on top of Solr search engine. On top of that, migration to newer version of Solr will make it safer.

GitHub Repository

GitHub : iANN_Solr