Lexicons for Sentiment, Affect, and Connotation

23.4 Semi-supervised induction of affect lexicons

1. What is the basic idea of semi-supervised lexicon induction?
   Start from seed words defining two poles of a semantic axis (good vs bad) and label each word by its similarity to the two seed sets; two families are axis-based and graph-based.
   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 23, §23.4, p. 525
2. How can seed sets handle domain differences in affect?
   Either start with one large seed lexicon and let induction adapt it to the domain, or choose different seeds per genre (e.g. separate seeds for general text, Twitter and finance, where profit, gains vs litigation, losses matter).
   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 23, §23.4, p. 525
3. Describe the semantic axis method for building a sentiment lexicon.
   Embed the seed words (off-the-shelf, trained on the domain, or fine-tuned); take each pole’s centroid \(V^+\), \(V^-\); the axis is \(V_{axis} = V^+ - V^-\); score each word by \(\cos(E(w), V_{axis})\).

   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 23, §23.4, pp. 525–526
4. How can a semantic axis score be turned into discrete lexicons?
   Apply a threshold (or similar method) to split the scored words into positive and negative lists.
   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 23, §23.4, p. 526
5. Describe the SentProp label propagation algorithm.
   (1) Build a graph connecting each word to its k nearest embedding neighbors, edges weighted by cosine-based similarity. (2) Pick positive and negative seeds. (3) Run random walks from each seed set; a word’s raw score is the probability of landing on it. (4) Combine: \(score^+(w) = \frac{raw^+(w)}{raw^+(w) + raw^-(w)}\). (5) Bootstrap over seed subsets for confidence.

   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 23, §23.4, pp. 526–527
6. How does Hatzivassiloglou and McKeown’s algorithm decide that two adjectives share polarity?
   Syntactic cues: adjectives frequently conjoined by and tend to share polarity (fair and legitimate, corrupt and brutal); those conjoined by but tend to be opposite (fair but brutal).
   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 23, §23.4, p. 527
7. What morphological cue signals opposite polarity?
   Morphological negation: adjectives with the same root differing by un-, im-, -less (adequate/inadequate, thoughtful/thoughtless) tend to have opposite polarity.
   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 23, §23.4, p. 528
8. How can WordNet expand a seed sentiment lexicon?
   Iteratively: Lex⁺ gets synonyms of positive words and antonyms of negative words; Lex⁻ gets synonyms of negative words and antonyms of positive words.
   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 23, §23.4, p. 528
9. What is SentiWordNet?
   Polarity assigned to WordNet synsets rather than words: each sense gets positive, negative and objective scores (summing to 1), e.g. estimable “deserving of esteem” is positive, “may be computed” purely objective. A classifier on glosses plus a random walk assigns the scores.
   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 23, §23.4, p. 528
10. Summarize the similarity metrics used by semi-supervised lexicon induction.
    Embedding cosine, coordination (and/but), morphology, and thesaurus structure, scoring words by similarity to positive seeds and dissimilarity to negative ones.
    • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 23, §23.4, p. 528