VL12 NL1 §

Question §

Notes §

Flashcards §

What is 🧑 BERT?#card §

= a 🧠 Machine Learning technique for ⚙ Natural Language Processing, which uses sparse, contextu^1642441138524 al embeddings.

Define 👀 Contextual embedding#card §

= 〰️ Word embedding that takes the context of a wor^1642441138535 d into account

Define 🤷‍♂️ Non-Contextual embedding#card §

= 〰️ Word embedding that does not take the context of a wor^1642441138546 d into account

What’s the 👔 Chomsky normal form?#card §

= a 📖 Formal grammar, which fulfills the follo^1642441138555 wing criteria:

no empty right-hand side every right-hand side has either 2x non-terminals or 1x terminal

What’s a 🏗 Formal generative grammar?#card §

= a fully explicit 📖 ^1642441138563 Formal grammar

What’s the formal definition of a 🏗 Formal generative grammar?#card §

• $N$: finite set of non-terminal symbols
• $\sum$: finite set of terminal symbols disjoint from $NL$
• $S$: start symbol, $\in$ $N$
• $P$: finite set of rules of type $\alpha \to \beta$
  • $\alpha ,\beta \in (N\bigcup \sum )$
  • $\alpha$ conaining min 1 non-ter^1642441138570 minal symbol

List 3/4 Phenomena#card §

• Subcategorization
  • verbs can take certain complements
  • but can also have number of optional modifiers
    • e.g. “I sleep here”, “I hardly ever sleep here”
• Agreement
  • main verb agrees with person and number
    • e.g. “We drive a car”, “He drives a car”
• Movement
  • certain constituents can be moved around
  • e.g. “He drove back”, “Back he drove”
• Coordination
  • phrases can be conjoined by means of and and or
  • “John and Ni^1642441138578 ck drove home”