Friday, 20 December 2019

Archaeology, Alexa and NLP

Hello everyone,
this post regards some test we are doing in these weeks about the application of NLP (Natural Language Processing) to archaeology. This research is conducted by our friend Andres Reyes (Arc-Team), an expert in this field.
Among the many possibilities of NLP in CH (Cultural Heritage), we decided to start with something particular and probably not so easy, but very useful for everyday work: a project manager for archaeology. The video below shows a preview of the system (how the system finds an old project).


To understand what I mean, I have to explain very fast why this tool would be a great help in our field. In Professional Archaeology (or, if you prefer Commercial Archaeology) projects can be divided in 4 main categories: excavations (probably the 70% of the work), surveys (and explorations in general), Cultural Heritage Enhancement (Valorization) and studies (mainly researches on specific archaeological and historical topics). From a logistical point of view, the most critical projects are the ones related with excavation and surveys, especially if performed in extreme conditions (Glacial Archaeology, High Mountain Archaeology, Underwater Archaeology, Speleoarchaeology, etc...), since in most cases the office (and all its comforts) is far away. Even if assisted by the strong computerization of the last 15 - 20 years, field operations can end up with errors, especially if many people work simultaneously to the same project from different area (for instance, a common mistake is giving the same code to different layers or artefacts).
A way to try to avoid errors is to use DBMSs (DataBase Management Systems) and GIS directly on the field, but this solution has some weakness, mainly related with the devices on which these software runs and with the necessity to find a comfortable location to insert the data (even if temporary). Thanks to the wider and wider coverage of internet and the new generation's smart-phones it is now simpler and faster to insert data into a main server trough a DBMS with a well designed interface (for GIS it is still better to work with a rugged laptop). Nevertheless these operations are still time consuming and keeps the archaeologist busy for a while, with all the difficulties coming from the use of a small touch-screen (gloves, dirty hands, rain, etc...). For this reason a Project Manager based on vocal commands could improve the work on the field, avoiding the main errors deriving by some of the most common stress factors (short time-table, weather conditions, several people working simultaneously, etc...).
Despite our decision to work with FLOSS, for this first experiment with NLP we decided to start with Amazon Alexa virtual assistant, for several reasons: the great effort of Amazon in developing the system, its strong diffusion among users and the good support in Italian (the language of our firts prototype). Nevertheless, as soon as we will have a first prototype, we plan to test and develop also open source solution, like Microft. BTW all our code will be released ASAP, with open source licenses, in this public repository on GitLab.
Currently our prototype is in a very early stage, but we already modified it a couple of time, with sensible changes in our strategy. For instance, in order to keep everything simple, at the beginning we based on shared google doc spreadsheets. This solution was more than enough to manage the list of codes related with US (Unità Stratigrafice, EN Stratigraphical Units), artefacts, samplings, documentations (in 3D and 2D), with also the possibility to keep controlled the budget and the working hours. soon we changed this strategy to have a more performing DBMS, based on the FLOSS PostgreSQL. Currently we are developing more options, like the possibility to ask to the Project Manager in which project we worked during a specific month.
I hope this post will be useful. If you want to collaborate to the project, please contact us. Have a nice day!

No comments:

Post a Comment

BlogItalia - La directory italiana dei blog Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 International License.