Geek Answers Handbook

How can I start building a wordnet for Turkish for use in mood analysis

Although I have background EE, I did not have the opportunity to attend natural language processing classes.

I would like to create a mood analysis tool for the Turkish language. I think it’s best to create a Turkish wordnet database, rather than translate the text into English and analyze it using transcoded text with the provided tools. ( is that? )

So what do you guys recommend me? First of all, taking NLP lessons from an open class site? I really don't know where to start. Could you help me and perhaps provide me with a step-by-step guide? I know this is an academic project, but I'm interested in creating skills as a hobby in this area.

Thanks in advance.

+7
source share
1 answer

Here is the process I used before (creating semantic networks in Japanese, Chinese, German, and Arabic):

  • Collect at least two English-Turkish dictionaries. They should be independent, and not derived from each other. You can use Wikipedia to automatically create one of your dictionaries. If you need to publish your network, you may need open source dictionaries, licensing fees, or a lawyer.
  • Use these dictionaries to translate English Wordnet, creating a confidence rating for each synthesizer.
  • Keep those with strong confidence, manually approve or install them with a medium or low level of confidence.
  • Finish it manually

I expanded this in the “Automatic WordNet Translation” section of my 2008 article: http://dcook.org/mlsn/about/papers/nlp2008.MLSN_A_Multilingual_Semantic_Network.pdf

(For your stated goal of the Turkish mood dictionary, there are other approaches not related to the semantic network. For example, “Semantic analysis and intellectual research”, Bing Liu, is a good research overview, but the semantic network approach, IMHO, always gives the best results in the long term and has so many other uses.)

+3
source

More articles:


All Articles