Scribd Logo
Upload

Welcome to Scribd!

  • 38 views

    Foundations of Natural Language Processing Wrapup, Review, and Exam Information

    Uploaded by

    pbtai
    This document provides a summary of the final lecture for the Foundations of Natural Language Processing course. It discusses [1] topics that have changed from previous years and those that remain the same, [2] the format of the upcoming exam, and [3] advice for preparing for the exam. The lecture reviews the course material and exams to help students focus their preparation.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Foundations of Natural Language Processing Wrapup, Review, and Exam Information

    Uploaded by

    pbtai
    38 views12 pages
    This document provides a summary of the final lecture for the Foundations of Natural Language Processing course. It discusses [1] topics that have changed from previous years and those that remain the same, [2] the format of the upcoming exam, and [3] advice for preparing for the exam. The lecture reviews the course material and exams to help students focus their preparation.

    Original Title

    18_slides

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document provides a summary of the final lecture for the Foundations of Natural Language Processing course. It discusses [1] topics that have changed from previous years and those that remain the same, [2] the format of the upcoming exam, and [3] advice for preparing for the exam. The lecture reviews the course material and exams to help students focus their preparation.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    38 views12 pages

    Foundations of Natural Language Processing Wrapup, Review, and Exam Information

    Uploaded by

    pbtai
    This document provides a summary of the final lecture for the Foundations of Natural Language Processing course. It discusses [1] topics that have changed from previous years and those that remain the same, [2] the format of the upcoming exam, and [3] advice for preparing for the exam. The lecture reviews the course material and exams to help students focus their preparation.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    You are on page 1of 12

    Foundations of Natural Language Processing

    Lecture 18
    Wrapup, review, and exam information
    Alex Lascarides

    20 March 2020

    Alex Lascarides FNLP Lecture 18 20 March 2020


    WARNING: this isn’t the same course it was in
    2015 and before
    When revising for the exam, past exam papers are useful, but be careful of
    overfitting.

    • Most topics in common with last year

    • Changed a lot in 2016

    • Different topics; some new approaches/models

    Alex Lascarides FNLP Lecture 18 1


    Topics in common with previous years

    • Corpora, annotation, evaluation


    • Ambiguity at all levels
    • N-gram models, entropy, smoothing
    • Noisy channel framework
    • Spelling correction, edit distance
    • HMMs, part-of-speech tagging
    • Syntax, parsing algorithms, PCFGs, other grammar formalisms
    • Lexical semantics: word senses

    Alex Lascarides FNLP Lecture 18 2


    Eliminated from previous years
    You will not be expected to answer questions about these topics.

    • corpus markup
    • mathematical details of backoff in N-gram models
    • details of forward-backward algorithm for HMMs
    • feature structure grammars
    • crowdsourcing in detail
    • implementation details of Good-Turing smoothing
    • pronoun resolution
    • machine translation

    Alex Lascarides FNLP Lecture 18 3


    New since 2015
    So past papers are not a good guide for these!

    • Updated discussion of evaluation


    • High-level overview of more modern smoothing methods (K-N)
    • More complete example of spelling correction (end-to-end system)
    • Generalized discussion of EM (showing application in both spelling correction
    and HMMs)
    • Text classification (tasks and methods)
    • Dependency grammar and related algorithms (e.g., SR parsing)
    • Semantic roles and distributional semantics

    Alex Lascarides FNLP Lecture 18 4


    Format of the exam
    As in previous years, the exam has two parts:

    • Part A: 8 short-answer questions, each worth 3 marks (total of 24 marks).

    • Part B: 3 longer questions worth 13 marks each, of which you must answer
    two (total of 26 marks).
    – Be clear which questions you are answering.
    – If you (start to) answer more than two, you must clearly cross out one
    answer.
    – Part A and each question for part B in separate booklets (i.e., 3 booklets
    overall).

    Alex Lascarides FNLP Lecture 18 5


    What counts and what doesn’t
    Things that do matter (not necessarily a complete list):

    • Complete answer (double check you’ve answered everything that was asked!)

    • Clear explanations/reasoning where appropriate

    • Correct equations, all variables defined

    • Legible

    Alex Lascarides FNLP Lecture 18 6


    What counts and what doesn’t
    Things that do not matter:

    • Perfect spelling/grammar/handwriting: as long as it is clear what you mean.


    Do not waste time writing drafts/copying over, but clearly cross out any
    scratch work that should not be marked. You can lose marks for have both
    correct and incorrect answers unless one is crossed out.

    • Full sentences. If a word or short phrase conveys the meaning, no need for
    more.

    Alex Lascarides FNLP Lecture 18 7


    Other ways to prepare

    • Lecture summary slides are a good place to start: they don’t have all the
    details, but make sure you understand the details underlying the main points
    mentioned.
    • Do the labs! Make sure you understand the answers you get
    • Heed any feedback on your courseworks and talk to your classmates or post
    on Piazza if you still don’t understand.
    • Post questions on Piazza. We will not always answer immediately but will
    try to ensure questions are answered. Exception: we will not answer any
    questions asked less than 48 hours before the exam.

    Alex Lascarides FNLP Lecture 18 8


    What courses follow on next year?

    • IAML: if you haven’t already taken it, do! ML underlies most of NLP, and
    fourth year courses assume a strong background.
    • Natural Language Understanding, Generation and MT (NLU+): more
    advanced models and algorithms for processing syntax, semantics, and
    discourse.
    • Text Technologies: focus on search engines, IR, working systems.
    • Automatic Speech Recognition: builds on knowledge from this course, but
    focuses on speech processing.

    Alex Lascarides FNLP Lecture 18 9


    Other related courses

    • Other machine learning courses (MLPR, MLP, PMR): These cover modern
    statistical approaches and deep learning models that are increasingly popular
    in NLP.

    • Extreme computing: for dealing with huge data sets.

    • Computational Cognitive Science, Topics in Cognitive Modeling: include


    sections on computational models of human language processing.

    Alex Lascarides FNLP Lecture 18 10


    That’s all folks!

    Alex Lascarides FNLP Lecture 18 11

    You might also like

    pFad - Phonifier reborn

    Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

    Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


    Alternative Proxies:

    Alternative Proxy

    pFad Proxy

    pFad v3 Proxy

    pFad v4 Proxy