Versions Compared

Old Version 1

changes.mady.by.user Rune Sætre

Saved on

compared with

New Version Current

changes.mady.by.user Rune Sætre

Saved on

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

Master Thesis Task Description:

Se endelig rapport i DAIM

Working title: Expanding an NLP-Based Bus Query System to Increase Its Usefulness

The task will be to expand the functionality of the natural language query system BusTUC, to include support for additional information deemed useful by potential users.

The autumn project leading up to this project shall be treated as the groundwork for this thesis. It identified suitable areas of expansion, at least some of which shall be implemented. For example, chronological bus stop lists and support for queries about bus frequencies were found to be popular potential improvements. Written dialogue support is also desirable.

The result shall be compared to BusTUC as it was before the work, and the significance of the work shall be evaluated.

 

Autumn Project Task Description

Vil integrere BusTUC med kart, sanntids-info, GPS og annet, for å finne ut hvordan ruteopplysninger best kan distribueres.

...

:

What is the most natural way to get information about bus scedules or other well organised and structured data?

The task will be to

...

extend the functionality of the spoken natural language query system to include support for some other useful and possibly relevant information.
Such information could be train routes (NSB), other train information, or possibly opening hours of for example Trondheim Torg or Studentersamfundet.
If opening hours for popular destinations are included, one might extend the system to answer questions such as "first bus to Trondheim Torg" better: that is, the first bus that will drop you off by an open Trondheim Torg.

One should look at different ways to include this functionality, as well as the consequences for the system as a whole. This could mean investigating user behavior and service quality. For example, what do users think of the system's usefulness and user-friendliness (with this extension)? And could one type of question be easily confused with another and damage the usefulness of the system?
With a prototype in place one could then investigate the actual consequences, and see if any parts of the system could be changed to provide better service overall.

(

...

The main focus will be on telephone directories (local/national), but other topics may be route information, cinemas, telephones, addresses, tourist information, news, football results etc. The motivation may be to answer any question that you would ask for intance to 1881."

The work can be done together with the FUIROS project, and new modes like Speech, GPS, Maps, and Real-time data integration and route updates could also be investigated.

Some existing work to build on can be found here

...

 

Siden mye av systemet allerede eksisterer, kan man nok omgjøre oppgaven til å bygge ut det som trengs av videre funksjonalitet, så lenge det er rimelig AI-relevant.

Dersom serveren fortsatt er oppe og fungerer, går det an å ringe den og se hva man får. Basert på dette kan man nok få en noe bedre oversikt.

...

)