Reading Suggestion: Adjectives and adverbs as indicators of affective language for automatic genre detection (2008)

Rittman, Robert and Nina Wacholder. (2008). Adjectives and adverbs as indicators of affective language for automatic genre detection. Proceedings of AISB 2008 Convention, Symposium on Affective Language. Aberdeen, Scotland, April 1-2, 2008.

Abstract. We report the results of a systematic study of the feasibility of automatically classifying documents by genre using adjectives and adverbs as indicators of affective language. In addition to the class of adjectives and adverbs, we focus on two specific subsets of adjectives and adverbs: (1) trait adjectives, used by psychologists to assess human personality traits, and (2) speaker-oriented adverbs, studied by linguists as markers of narrator attitude. We report the results of our machine learning experiments using Accuracy Gain, a measure more rigorous than the standard measure of Accuracy. We find that it is possible to classify documents automatically by genre using only these subsets of adjectives and adverbs as discriminating features. In many cases results are superior to using the count of (a) nouns, verbs, or punctuation, or (b) adjectives and adverbs in general. In addition, we find that relatively few speaker-oriented adverbs are needed in the discriminant models. We conclude that at least in these two cases, the psychological and linguistic literature leads to identification of features that are quite useful for genre detection and for other applications in which identification of style and other non-topical characteristics of documents is important.

Full paperProceedings of the AISB 2008 Symposium on Affective Language in Human and Machine

AISB 2008 Convention: Communication, Interaction and Social Intelligence
1st-4th April 2008 University of Aberdeen

Related Posts

Additional References

2 comments for “Reading Suggestion: Adjectives and adverbs as indicators of affective language for automatic genre detection (2008)

    25 February, 2012 at 07:15

    From LinkedIn: The WebGenre R&D Group
    See Discussion

    Glenn Mungra, Kei Kurakawa like this

    Jan Jasik • Marina, … reading prof. Goswami, David Wilcock, etc… none of that comes to mind and holding on to the semantics of EA as a loyal proponent; the ontology should help when parsing through semantic universe. Grammar could be an expression of taxonomy and as such extend ontology as much as it’d lend itself. ‘Affective language’ could be just a desire helping us to feel comfortable to follow others. Trusting anything or anyone now becomes increasingly a challenge.

    Marina Santini • Jan, it seems to me that the web has unveiled the universe of strong emotions, public likes and dislikes. Language on the web reflects this affective dimension…

    Jan Jasik • Marina, web is compensating us for diminishing of traditional space we used to have to debate, make mistakes and learn in the process. There has always been trend to sanitize public forum, but motives have not been innocent at all… namely control of our consciousness. We may need meta-adjectives and meta-adverbs projecting our intentions more accurately due to the lack of proper attributes?

    Marina Santini • That’s an idea 🙂

    Jan Jasik • Sure. One may wander if it meant: 8-), |-), !-), 8^), |^), !^)… or something else not revealing intentions or taking consciousness under consideration (covering only one geo-political community). A lot of potential for misunderstandings … (8-)

    25 February, 2012 at 07:23

    From LinkedIn: UTMA – Ubiquitous Text Mining and Analytics Group

    See Discussion

    Ina Lauth likes this

    Menno Mafait • Without a fundamental approach to AI / NLP / Semantics, information systems will never be intelligent. They only will be mimicking / simulating intelligence.

    Allow me to illustrate by a metaphor: The current approach to AI / NLP / Semantics is like building a flight simulator, while my system ( is the real thing: an airplane to moves from A to B.

    Cause: The current approach to AI / NLP / Semantics has no proven foundation in intelligence.

    Marina Santini • Hi Menno, your website (Thinknowlogy, Fundamentally designed Artificial Intelligence) is exciting.

    As far as intelligence is concerned, well you might have in mind some kind of logic /IQ intelligence that, at least for humans, has been proved to be insufficient to cope with real life and manage emotional/social interactions. What I have in mind right now is the view puts forward by David Goleman in “Emotional Intelligence – Why it Can Matter More Than IQ”.

    My personal challenge and vision focus towards information systems that contain sparks of emotional and social intelligence and that can learn how to develop them more and more through interaction with humans. Handling emotions, sentiments and attitudes is a good way to start, then…

    Menno Mafait • Marina, a lot of development is wasted on simulating intelligence, simply because AI scientists don’t have a clue what intelligence really is: There is not even scientific consensus about the definition for intelligence. How can we create it then?

    About “has been proved to be insufficient to cope with real life”: If all effort spend on simulating intelligence is used to create a proven foundation in intelligence, information systems are able to cope with real life even better than simulated intelligence.

    The same goes for Emotional Intelligence, which is simulated too.

    Marina Santini • Not sure about the meaning of “simulated intelligence” or “real intelligence”. They might be new categories… Logic is not more satisfying or more real than emotions. Just the opposite. I always think of Wittgenstein’s experience with his Tractatus when I come across extensive reliance on logic…

    Anyway, I believe it is good to propose a range of different approaches (real or simulated, in your own terms) and let the users decide what they feel more comfortable with 🙂

    Have a nice day, Menno.

    Menno Mafait • Marina, no AI solution – other then mine – is proven to be based on intelligence.
    An example:

    • IBM’s Watson is able to find a needle in the haystack of unstructured texts. But Watson is only engineered around the problem of intelligence. The real problem is: structuring texts. My system is able to structure texts. It is like browsing a library.
    However, my fundamental research has a major – but temporary – drawback: It takes a lot of time and effort to mature to develop its gray and boring foundation. It is still in its infancy and only accepts sentences with a very limited grammar.

    I am still the only one in the world developing a fundamental solution:
    • Since my system invalidates the non-fundamental approach of science, I am excluded from any scientific and European funding;
    • And despite my public analysis and design papers (see website), nobody else is able to understand the beauty / simplicity of intelligence.

Leave a Reply

Your email address will not be published. Required fields are marked *