AMI provides a generic infrastructure where plugins can search, index or transform structured documents on a high-through basis. The typical input is structured, normalized, tagged XHTML, possibly containing (or linked to) SVG and PNG files. The plugins are designed to analyse text or graphics or a combination according to the discipline.

There are binaries provided with releases which enable running of ami's plugins; shell and batch scripts are provided to make running these easier. Detailed tutorials of these can be found in the ContentMine software tutorials.

Run on a CTree which contains scholarly.html files of the papers you are analysing. This can be made using Norma.

Schema

ami2-<pluginname> --project <foldername> <plugin option> <options relating to plugin>
ami2-gene --project zika --g.gene --g.type human

mvn package

After the build, shell scripts and batch files to run plugins are in target/appassembler/bin/. Further documentation is in docs/.

AMI has a plugin architecture where each problem or community has its own plugin. Examples are "species", "sequence", "regex" and soon some chemistry.

It is often straightforward to develop text-based searches, and this is accessible to most committed scientists. Graphics is always harder and requires bespoke programming. Plugins have been developed for at least:

• indexing text by regular expressions (regex).
• Genbank IDs
• PDB ids
• farm-related / agronomy terms
• chemical species (OSCAR)
• computational phylogenetics
• terms identifying Ebola and other haemorrhagic diseases

• phylogenetic trees
• chemical structures

The plugin architecture is moderately stable and it requires very little alteration to the codebase to add a new one (hopefully soon this can be done automatically by configuration files).

The input must be a QuickscrapeNorma directory (QSNorma). This must contain scholarly.html which is used for analysis. When a plugin is run, the output is to the results directory, wiwth a subdirectory for each plugin and a sub-subdirectory for each plugin option: Example:

http_www.foo_1_2/
    fulltext.xml
    fulltext.pdf
    scholarly.html
    results/
        words/
            frequency/
                results.xml
            lengths/
                results.xml
        regex/
            consort0/
                results.xml
            publication/
                results.xml