CPGP Concept matching calculation with inner product

Created: 2014-05-27 11:49:38 Last updated: 2014-06-30 09:24:31

Download Workflow

Requirements:

Have MySQL installed with a database called mydb or please create one. Have a mysql connector file. This can be downloaded from mysql.com. Rename the downloaded file to mysql-connector-java.jar and place it in the the local lib file of Taverna as written on the dependencies tab. Choose this dependency by rightclicking a beanshell script and selecting it for the whole workflow as system classloader.

Download the Peregrine SKOS CLI from: https://trac.nbic.nl/biosemantics/downloads Install LVG2013Lite. See: https://trac.nbic.nl/biosemantics/wiki/Peregrine%20SKOS%20CLI

Change lvg.properties LVG_DIR=/home/path/to/lvg2013lite/ This file can be found in:

/home/path/to/lvg2013lite/data/config/lvg.properties

Copy properties file from production.properties which can be found on: https://trac.nbic.nl/biosemantics/wiki/Peregrine%20SKOS%20CLI

Change the normalizer.lvg.properties and normalizer.lvg.binaryCache properties to point to the LVG installation path. This file can be obtained from https://trac.nbic.nl/biosemantics/wiki/Peregrine%20SKOS%20CLI

Have Peregrine indexer installed locally and give it the correct path in the component named "Indexer_tool". Which can be found in the nested workflow named Peregrine indexer.

If using a Windows machine please install cygwin. Set cygwin in the $PATH of the environment variabale. Go to taverna File --> preferences --> Tool invocation --> set modify on explicit locations --> edit a default local to Shell: C:\Windows\system32\cmd.exe /c and save that. -->Set Modify on: symbolic locations --> edit --> choose default local and save. When using a unix environment make sure the default local is set Shell into /bin/sh -c. Please make sure this is also set into the indexer tool part.

Have JAVA_HOME C:\Program Files\jdk 1.7.0 in environment variabale when using Windows.

The nested workflow Get_abstract uses a sql command to retrieve abstracts which assumes are located in the database mydb. If this is not the case, please change the database in the SQL value.

Preview

Download as scalable diagram (SVG)

Run

Run this Workflow in the Taverna Workbench...

Option 1:

Copy and paste this link into File > 'Open workflow location...'
http://myexperiment.org/workflows/4300/download?version=2
[ More Info Expand ]

Workflow Components

Authors (1)

Titles (1)

Descriptions (1)

Requirements: Have MySQL installed with a database called mydb or please create one. Have a mysql connector file. This can be downloaded from mysql.com. Rename the downloaded file to mysql-connector-java.jar and place it in the the local lib file of Taverna as written on the dependencies tab. Choose this dependency by rightclicking a beanshell script and selecting it for the whole workflow as system classloader. Download the Peregrine SKOS CLI from: https://trac.nbic.nl/biosemantics/downloads Install LVG2013Lite. See: https://trac.nbic.nl/biosemantics/wiki/Peregrine%20SKOS%20CLI Change lvg.properties LVG_DIR=/home/path/to/lvg2013lite/ This file can be found in: /home/path/to/lvg2013lite/data/config/lvg.properties Copy properties file from production.properties which can be found on: https://trac.nbic.nl/biosemantics/wiki/Peregrine%20SKOS%20CLI Change the normalizer.lvg.properties and normalizer.lvg.binaryCache properties to point to the LVG installation path. This file can be obtained from https://trac.nbic.nl/biosemantics/wiki/Peregrine%20SKOS%20CLI Have Peregrine indexer installed locally and give it the correct path in the component named "Indexer_tool". Which can be found in the nested workflow named Peregrine indexer. If using a Windows machine please install cygwin. Set cygwin in the $PATH of the environment variabale. Go to taverna File --> preferences --> Tool invocation --> set modify on explicit locations --> edit a default local to Shell: C:\Windows\system32\cmd.exe /c and save that. -->Set Modify on: symbolic locations --> edit --> choose default local and save. When using a unix environment make sure the default local is set Shell into /bin/sh -c. Please make sure this is also set into the indexer tool part. Have JAVA_HOME C:\Program Files\jdk 1.7.0 in environment variabale when using Windows. The nested workflow Get_abstract uses a sql command to retrieve abstracts which assumes are located in the database mydb. If this is not the case, please change the database in the SQL value.

Dependencies (1)

Inputs (1)

Name	Description
db_properties	Please load a file in as seen in the example where the user, password, url (which is the database it should use), and driver is stated.

Processors (4)

Name	Type	Description
Calculate_Contingency_Table	workflow	A nested workflow that retrieves the values of a 2x2 contingency table and puts the co-occurrence of made concept pairs into the SQL table TableName_co_occurrence. This is done by creating concept pairs which is retrieved from the Sql table TableName_occurrence. This contingency table consits out of a matrix where two concepts co-occur, concept A only occurs and not concept B, concept B occurs and not concept A and neither concepts occur in literature. Example of a 2x2 contingency table of concept pair A B: B Not B A AB A not A B not A not B This is calculated for every possible concept pair.
Calculate_Inner_Product	workflow	A nested workflow to calculate the inner product between concept profile pairs for concept profile matching. The calculated values are afterwards inserted into the inner product SQL table. Example of calculating the inner product between two concept profiles: A B \| \| \|__UC value__A x A__UC value__\| \| + \| \|__UC value__B x B__UC value__\| \| \| \|__UC value__C D__UC value__\| The inner product of the two concept profile A and B is the common concept multplied between each other and added: (UC value AA x UC value BA) + (UC value AB x UC value BB) = inner product
timestamp_value	stringconstant	Value alpha
table_name_value	stringconstant	Value alpha

Beanshells (4)

Name	Description	Inputs	Outputs
Beanshell_generate_SQL_statement_Get_distinct_concept	A Beanshell script that retrieves every concept from the Sql table TableName_occurrence.	table_name	out1
Beanshell_make_concept_pairs	A Beanshell script that gets as input a list of concepts to create concept pairs.	concept_B concept_A	concept_pair
Beanshell_SQL_generate_Get_Inner_Product	A Beanshell script that generates a SQL statement to calculate the inner product between two concept profiles. These concept profiles only have concepts that both have in common. The concepts that are in common in the concept profiles are multiplied with each others UC value and added. This summed up value of two concept profiles is the inner product. For example: Concept profile A, has the concept A, B, C and D. Concept profile B has the concepts B, C, E. This means the common concepts for both concept profiles are B, C. These values are multiplied and added. This final value is the inner product between two concept profiles.	concept_pairs timestamp	out1
Beanshell_SQL_generate_INSERT_INTO_Inner_Product_table	A Beanshell script that generates a SQL statement to insert the inner product values calculated of two concept profiles into the inner product table.	concept_pairs timestamp inner_product_value	out1

Outputs (1)

Name	Description
end

Datalinks (6)

Source	Sink
db_properties	Calculate_Contingency_Table:propertyString
table_name_value:value	Calculate_Contingency_Table:table_name
Calculate_Contingency_Table:outputlist	Calculate_Inner_Product:concept_pairs
db_properties	Calculate_Inner_Product:propertyString
timestamp_value:value	Calculate_Inner_Product:timestamp
Calculate_Inner_Product:output	end