Page History
Scrollbar | ||
---|---|---|
|
...
Page info | ||||
---|---|---|---|---|
|
Section | |||||||||||||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
When first installed, LexEVS (an open-source, enterprise-wide terminology server) comes with no terminologies loaded into it. This documentation will cover the means for loading most content types that can be loaded. LexEVS was built to accommodate a wide variety of input and meld it into a common form - unifying many common source formats.
...
Most terminology loads can be easily accomplished by pointing either the LexEVS commands or the LexEVS administrative GUI at the terminology source file and running the loader. Before you try this, you should look to see if the terminology you want to load is a special case (at the bottom of this page). Generic loading instructions can be found for the LexEVS administrative GUI or the LexEVS loader commands. For many source formats you can use a variation of the following LexEVS command:
...
While terminologies are being loaded, you can monitor the progress using the LexEVS logs (both 'load' and 'full' log) and if using MySQL, use INNODB tools to monitor Inserts per second. (SHOW INNODB STATUS
Multiexcerpt include | ||||||||
---|---|---|---|---|---|---|---|---|
|
...
Some terminologies are special cases and need special handling. Each of these has its own documentation:
Special Case | How to handle | ||||||
---|---|---|---|---|---|---|---|
The NCI Thesaurus differs from other OWL formatted resources and as a result you should follow this documentation. | |||||||
OWL terminologies do not normally require special handling, but LexEVS offers some advanced loading options users may take advantage of. | |||||||
Terminologies in RRF format typically come from the National Library of Medicine's (NLM) Unified Medical Language System (UMLS). Many terminologies are a subset of the UMLS such as LOINC, SNOMED, MedDRA, HUGO, GO, and ICD to name a few. The terminology you're interested in is a subset of the UMLS if:
|
| |||||||
Loading Asserted Value Set Definitions and Indexes | Source Asserted Value Set Loads require that value set definitions based on source asserted value sets be loaded. Supporting Indexes for searches on these value sets also need a separate load from the source terminology load (At the moment this is focused on the NCI Thesaurus.) | ||||||
Loading the Graph Database into ArangoDb for a Given Terminology | LexEVS supports use of a graphing database, ArangoDb for high speed graph resolution of its relationships. These can be loaded from any terminology in the current LexEVS terminology service. The graphs are loaded from the table of associations persisted into the LexGrid data model in MySQL and can be accessed from either a REST graph service or a client service the NodeGraphResolutionService. It is meant to supplement, but not replace, the LexEVS CodedNodeGraph API. | ||||||
None of the above match and you cannot find a suitable source format. | Many terminology providers produce more than one source format that can be downloaded. Source formats such as text only, CSV, tab delimited, and spreadsheets are not acceptable source formats for LexEVS. If you can not find an acceptable source format for a terminology to load into LexEVS then one option is to download versions that have been placed on BioPortal
|
|
...
Common Errors When Loading
Error | Remedy or Indication |
---|---|
Out of Memory Error Heap Error Perm Gen Error | Generally memory related errors that indicate the heap space and/or perm gen space need to be increased when starting the Java VM. |
Data Truncation Error | Source formats change over the years and sometimes this results in this database related error. It indicates that a column size is too small for whatever element has been pulled from the source file. This will require an update to LexEVS to fix and a temporary workaround can be to edit the source so that the element is made short enough to fit the database column. |
Data Base Connection Error | Check to see if the DBMS is up and running, that your database exists, that the connection parameters are correct, and that proper privileges exist for the connecting user. This error may manifest itself as DAO related errors being generated by the spring framework at some levels of execution in the logs, including failing to create a dao list. |
Too Many Files Open | Linux system error requires setting of system properties. This happens when the Lucene index is large and is being reindexed on loading a new terminology or is being optimized or cleaned up. |
Loader Hangs with no Errors | This can happen when there is processor capacity is maxed out, or there is network latency of one variety or another. For very large terminologies it may be necessary to just wait this out, but this can be helped by working with local Lucene files (highly recommended), a local database, and moving load operations to a system where there are multiple processor cores and adequate memory (16GB and more). Even for loads that are memory efficient, such as those using the Spring Batch functions, indexing is still memory bound and can go much faster with more memory. |
Scrollbar | ||
---|---|---|
|