Semantic Frames to Predict Stock Price Movement
Xie, Boyi and Passonneau, Rebecca J. and Wu, Leon and Creamer, Germán G.

Article Structure

Abstract

Semantic frames are a rich linguistic resource.

Introduction

A growing literature evaluates the financial effects of media on the market (Tetlock, 2007; Engel-berg and Parsons, 2011).

Motivation

Financial news is a rich vein for NLP applications to mine.

Related Work

NLP has recently been applied to financial text for market analysis, primarily using bag-of-words (BOW) document representation.

Methods

We propose two approaches for the use of semantic frames.

Dataset

We use publicly available financial news from Reuters from January 2007 through August 2012.

Experiments

Our current experiments are carried out for each year, training on one year and testing on the next.

Discussion

7.1 Semantic Parse Quality

Conclusion

We have presented a model for predicting stock price movement from news.

Topics

bag-of-words

Appears in 8 sentences as: Bag-of-Words (2) bag-of-Words (2) bag-of-words (4)
In Semantic Frames to Predict Stock Price Movement
  1. Our main contribution is a novel tree representation based on semantic frame parses that performs significantly better than enriched bag-of-words vectors.
    Page 1, “Introduction”
  2. On the polarity task, the semantic frame features encoded as trees perform significantly better across years and sectors than bag-of-words vectors (BOW), and outperform BOW vectors enhanced with semantic frame features, and a supervised topic modeling approach.
    Page 2, “Introduction”
  3. Bag-of-Words (BOW) document representation is difficult to surpass for many document classification tasks, but cannot capture the degree of semantic similarity among these sentences.
    Page 2, “Motivation”
  4. NLP has recently been applied to financial text for market analysis, primarily using bag-of-words (BOW) document representation.
    Page 3, “Related Work”
  5. Table 1: FWD features (Frame, bag-of-Words , part-of-speech DAL score) and their value types.
    Page 3, “Related Work”
  6. Table 1 lists 24 types of features, including semantic Frame attributes, bag-of-Words , and scores for words in the Dictionary of Affect in Language by part of speech (pDAL).
    Page 4, “Methods”
  7. Bag-of-Words features include term frequency and tfidf of unigrams, bigrams, and trigrams.
    Page 4, “Methods”
  8. Experiments evaluate the FWD and SemTree feature spaces compared to two baselines: bag-of-words (BOW) and supervised latent Dirichlet allocation (sLDA) (Blei and McAuliffe, 2007).
    Page 6, “Experiments”

See all papers in Proc. ACL 2013 that mention bag-of-words.

See all papers in Proc. ACL that mention bag-of-words.

Back to top.

classification tasks

Appears in 7 sentences as: classification task (1) classification tasks (6)
In Semantic Frames to Predict Stock Price Movement
  1. Our experiments test multiple text representations on two binary classification tasks , change of price and polarity.
    Page 1, “Abstract”
  2. Our experiments test several document representations for two binary classification tasks , change of price and polarity.
    Page 1, “Introduction”
  3. Bag-of-Words (BOW) document representation is difficult to surpass for many document classification tasks , but cannot capture the degree of semantic similarity among these sentences.
    Page 2, “Motivation”
  4. Our work addresses classification tasks that have potential relevance to an influential financial model (Rydberg and Shephard, 2003).
    Page 3, “Related Work”
  5. Our two binary classification tasks for news, price change and polarity, are analogous to their activity and direction.
    Page 3, “Related Work”
  6. gested as one of the best methods to summarize into a single value the confusion matrix of a binary classification task (Jurman and Furlanello, 2010; Baldi et al., 2000).
    Page 7, “Experiments”
  7. Models are constructed using linear kernel support vector machines for both classification tasks .
    Page 7, “Experiments”

See all papers in Proc. ACL 2013 that mention classification tasks.

See all papers in Proc. ACL that mention classification tasks.

Back to top.

tree kernel

Appears in 6 sentences as: tree kernel (4) Tree Kernels (1) tree kernels (1)
In Semantic Frames to Predict Stock Price Movement
  1. We introduce a novel tree representation, and use it to train predictive models with tree kernels using support vector machines.
    Page 1, “Abstract”
  2. The second is a tree representation that encodes semantic frame features, and depends on tree kernel measures for support vector machine classification.
    Page 3, “Methods”
  3. The tree kernel (Moschitti, 2006; Collins and Duffy, 2002) is a function of tree similarity, based on common substructures (tree fragments).
    Page 4, “Methods”
  4. 3'SVM-light: http://svmlight.joachims.org and Tree Kernels in SVM-light: http://disi.unitn.it/moschitti/Tree-Kernel.htm.
    Page 7, “Experiments”
  5. To analyze which were the best performing features within sectors, we extracted the best performing frame fragments for the polarity task using a tree kernel feature engineering method presented in Pighin and Moschitti (2009).
    Page 8, “Discussion”
  6. Our semantic frame-based model benefits from tree kernel learning using support vector machines.
    Page 9, “Conclusion”

See all papers in Proc. ACL 2013 that mention tree kernel.

See all papers in Proc. ACL that mention tree kernel.

Back to top.

binary classification

Appears in 5 sentences as: binary classification (5)
In Semantic Frames to Predict Stock Price Movement
  1. Our experiments test multiple text representations on two binary classification tasks, change of price and polarity.
    Page 1, “Abstract”
  2. Our experiments test several document representations for two binary classification tasks, change of price and polarity.
    Page 1, “Introduction”
  3. Our two binary classification tasks for news, price change and polarity, are analogous to their activity and direction.
    Page 3, “Related Work”
  4. Both tasks are treated as binary classification problems.
    Page 6, “Experiments”
  5. gested as one of the best methods to summarize into a single value the confusion matrix of a binary classification task (Jurman and Furlanello, 2010; Baldi et al., 2000).
    Page 7, “Experiments”

See all papers in Proc. ACL 2013 that mention binary classification.

See all papers in Proc. ACL that mention binary classification.

Back to top.

feature space

Appears in 5 sentences as: Feature Space (1) feature space (2) feature spaces (2)
In Semantic Frames to Predict Stock Price Movement
  1. We explore a rich feature space that relies on frame semantic parsing.
    Page 3, “Related Work”
  2. 4.2 SemTree Feature Space and Kernels
    Page 4, “Methods”
  3. We propose SemTree as another feature space to encode semantic information in trees.
    Page 4, “Methods”
  4. Experiments evaluate the FWD and SemTree feature spaces compared to two baselines: bag-of-words (BOW) and supervised latent Dirichlet allocation (sLDA) (Blei and McAuliffe, 2007).
    Page 6, “Experiments”
  5. SVM-light with tree kernels3 (Joachims, 2006; Moschitti, 2006) is used for both the FWD and SemTree feature spaces .
    Page 7, “Experiments”

See all papers in Proc. ACL 2013 that mention feature space.

See all papers in Proc. ACL that mention feature space.

Back to top.

semantic parses

Appears in 5 sentences as: Semantic Parse (1) semantic parse (1) semantic parses (2) semantic parsing (2)
In Semantic Frames to Predict Stock Price Movement
  1. We explore a rich feature space that relies on frame semantic parsing .
    Page 3, “Related Work”
  2. The semantic parses of both methods are derived from SEMAFOR1 (Das and Smith, 2012; Das and Smith, 2011), which solves the semantic parsing problem by rule-based target identification, log-linear model based frame identification and frame element filling.
    Page 3, “Methods”
  3. The top of Figure 2 shows the semantic parse for sentence a from section 2; we use it to illustrate tree construction for designated object Oracle.
    Page 4, “Methods”
  4. 7.1 Semantic Parse Quality
    Page 7, “Discussion”
  5. On a small, randomly selected sample of sentences from all three sectors, two of the authors working independently evaluated the semantic parses , with approximately 80% agreement.
    Page 7, “Discussion”

See all papers in Proc. ACL 2013 that mention semantic parses.

See all papers in Proc. ACL that mention semantic parses.

Back to top.

part-of-speech

Appears in 4 sentences as: part-of-speech (4)
In Semantic Frames to Predict Stock Price Movement
  1. Table 1: FWD features (Frame, bag-of-Words, part-of-speech DAL score) and their value types.
    Page 3, “Related Work”
  2. (2009) introduced part-of-speech specific DAL features for sentiment analysis.
    Page 4, “Methods”
  3. We remove stop words and use Stanford CoreNLP for part-of-speech tagging and named entity recognition.
    Page 7, “Experiments”
  4. We proposed FWD (Frames, BOW, and part-of-speech specific DAL) features and SemTree data representations.
    Page 9, “Conclusion”

See all papers in Proc. ACL 2013 that mention part-of-speech.

See all papers in Proc. ACL that mention part-of-speech.

Back to top.

support vector

Appears in 4 sentences as: support vector (4)
In Semantic Frames to Predict Stock Price Movement
  1. We introduce a novel tree representation, and use it to train predictive models with tree kernels using support vector machines.
    Page 1, “Abstract”
  2. The second is a tree representation that encodes semantic frame features, and depends on tree kernel measures for support vector machine classification.
    Page 3, “Methods”
  3. Models are constructed using linear kernel support vector machines for both classification tasks.
    Page 7, “Experiments”
  4. Our semantic frame-based model benefits from tree kernel learning using support vector machines.
    Page 9, “Conclusion”

See all papers in Proc. ACL 2013 that mention support vector.

See all papers in Proc. ACL that mention support vector.

Back to top.

semantic roles

Appears in 3 sentences as: Semantic role (1) semantic role (1) semantic roles (2)
In Semantic Frames to Predict Stock Price Movement
  1. Semantic role labeling using FrameNet has been used to identify an opinion with its holder and topic (Kim and Hovy, 2006).
    Page 3, “Related Work”
  2. FrameNet defines hundreds of frames, each of which represents a scenario associated with semantic roles , or frame elements, that serve as participants in the scenario the frame signifies.
    Page 4, “Methods”
  3. SemTree features capture the differences between semantic roles for the same frame, and between the same semantic role in different frames.
    Page 8, “Discussion”

See all papers in Proc. ACL 2013 that mention semantic roles.

See all papers in Proc. ACL that mention semantic roles.

Back to top.

sentiment analysis

Appears in 3 sentences as: Sentiment analysis (1) sentiment analysis (2)
In Semantic Frames to Predict Stock Price Movement
  1. Sentiment analysis figures strongly in NLP work on news.
    Page 3, “Related Work”
  2. For deep representation of sentiment analysis , Ruppenhofer and Rehbein (2012) propose SentiFrameNet.
    Page 3, “Related Work”
  3. (2009) introduced part-of-speech specific DAL features for sentiment analysis .
    Page 4, “Methods”

See all papers in Proc. ACL 2013 that mention sentiment analysis.

See all papers in Proc. ACL that mention sentiment analysis.

Back to top.

PDF available from www.aclweb.org.
Send us an email