Further idea tweaking

docs/paper.md

@@ -26,11 +26,11 @@ bibliography: paper.bib
 
 `scores` is a Python package containing mathematical functions for the verification, evaluation and optimisation of forecasts, predictions or models. It primarily supports the geoscience and earth system science communities. It also has wide potential application in machine learning, and domains other than meteorology, geoscience and weather. 
 
-`scores` is focused on supporting xarray [@Hoyer:2017] datatypes for earth system data. It also aims to be compatible with pandas and geopandas, and to work with NetCDF4, hdf5, Zarr and GRIB data sources among others. Scores is designed to utilise Dask for scaling and performance.
+`scores` includes novel scores not commonly found elsewhere (e.g. FIRM, FlipFlop Index), complex scores (e.g. threshold weighted CRPS), more common scores (e.g. MAE, RMSE) and statistical tests (such as the Diebold Mariano test). 
 
 All of the scores and metrics in this package have undergone a thorough statistical and scientific review. Every score has a companion Jupyter Notebook tutorial demonstrating its use in practice.
 
-At the time of writing, the scores contained in this package are: Mean Squared Error (MSE), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), the Fixed Risk Multicategorical (FIRM) score [@Taggart:2022a], Continuous Ranked Probability Score (CRPS) for Cumulative Distribution Functions (CDFs) (including threshold-weighting, see [@Gneiting:2011]), the FlipFlop index [@Griffiths:2019; @griffiths2021circular], Receiver Operating Characteristic (ROC) curves, the quantile score, and the Murphy score [@Ehm:2016]. It also includes the Diebold-Mariano statistical test [@Diebold:1995] with both the [@Harvey:1997] and [@Hering:2011] modifications. Additionally it contains isotonic regression which is becoming an increasingly important tool in forecast verification and can be used to generate stable reliability diagrams [@dimitriadis2021stable].
+`scores` is focused on supporting xarray [@Hoyer:2017] datatypes for earth system data. It also aims to be compatible with pandas and geopandas, and to work with NetCDF4, hdf5, Zarr and GRIB data sources among others. Scores is designed to utilise Dask for scaling and performance.
 
 The software repository can be found at [https://github.com/nci/scores/](https://github.com/nci/scores/).
 
@@ -40,14 +40,16 @@ The research purpose of this software is (a) to mathematically verify and valida
 
 ## Key Benefits
 
-`scores` includes novel scores not commonly found elsewhere (e.g. FIRM, FlipFlop index), complex scores (e.g. threshold weighted CRPS), more common scores (e.g. MAE, RMSE) and statistical tests (such as the Diebold Mariano test). Scores provides its own implementations where relevant to avoid extensive dependencies. 
-
 `scores` works with n-dimensional data (e.g., geospatial, vertical and temporal dimensions) for both point-based and gridded data. It has proper treatments for missing data, masking of data and weighting of results.
 
+`scores` avoids extensive dependencies by providing its own implementations where relevant.
+
 `scores` was designed to work effectively with the libraries, data structures and methods commonly in use for scoring, verifying and evaluating earth system models including Numerical Weather Prediction (NWP) models, forecasts for specific sites and weather phenomena such as thunderstorms. It can effectively handle the dimensionality, data size and requirements of the modelling community.
 
 `scores` is highly modular and has a minimal set of requirements. It is intended to be easy to integrate and utilise in a wide variety of environments. It has been tested and used on workstations, servers and in high performance computing (supercomputing) environments. 
 
+At the time of writing, the scores contained in this package are: Mean Squared Error (MSE), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), the Fixed Risk Multicategorical (FIRM) score [@Taggart:2022a], Continuous Ranked Probability Score (CRPS) for Cumulative Distribution Functions (CDFs) (including threshold-weighting, see [@Gneiting:2011]), the FlipFlop Index [@Griffiths:2019; @griffiths2021circular], Receiver Operating Characteristic (ROC) curves, the quantile score, and the Murphy score [@Ehm:2016]. It also includes the Diebold-Mariano statistical test [@Diebold:1995] with both the [@Harvey:1997] and [@Hering:2011] modifications. Additionally it contains isotonic regression which is becoming an increasingly important tool in forecast verification and can be used to generate stable reliability diagrams [@dimitriadis2021stable].
+
 ## Roadmap and Future Development
 
 The `scores` roadmap includes support for machine learning library integration, further optimisation of existing scores and the addition of more scoring functions for verification purposes.