Hi Laurent, I'm kind of doing something similar. There's a concise WordNet library called RiTa. Wordnet is a large lexical database of english words. With RiTa, you don't even have to install the database, you can just call the API to give you random words and do ontological things like:
* Synonyms, Antonyms, Hypernyms, & Hyponyms * Holonyms, Meronyms, Coordinates, & Similars * Nominalizations, Verb-Groups, & Derived-terms * Glosses, See-Alsos, Examples, & Descriptions * Support for Pattern-matching, Soundex, Anagrams, etc. Docs: http://www.rednoise.org/rita/wordnet/documentation/index.htm Tutorial: http://www.shiffman.net/teaching/a2z/wordnet/ I couldn't find maven repositories, so I made my own. <dependency> <groupId>org.rednoise</groupId> <artifactId>RiTaWN</artifactId> <version>024</version> </dependency> <dependency> <groupId>org.rednoise</groupId> <artifactId>supportWN</artifactId> <version>024</version> </dependency> <dependency> <groupId>org.processing</groupId> <artifactId>core</artifactId> <version>1.0</version> </dependency> <repository> <id>stavish-public-repository</id> <name>Publicly available Maven 2 repository by Todd Stavish</name> <url>http://www.stavi.sh/repository/maven2/</url> <snapshots> <enabled>true</enabled> </snapshots> </repository> Regards, Todd On Sun, Nov 22, 2009 at 8:53 AM, Laurent Laborde <[email protected]> wrote: > > Hi ! > > For a fun side-project i want to graph a markov chain (a Markov Text > Generator). > > 1 node = 1 word. > > When, while parsing a text, i find a word followed by another word i > create a "node-word" if it doesn't exist, and create relationship. > if the word already exist : i create an outgoing relationship between > the 2 words. > if the relationship already exist i just increment a counter (this > word was followed by the next word one more time). > > Now i need to store the probability of a specific word being followed > by another word. > including a word looping to itself. > > If you're not familiar with markov chain, a simple picture should > explain it : > http://www.leda-tutorial.org/en/unofficial/Pictures/MarkovChain.png > > Storing the probabilty of a link to be followed should be ideally > stored in the relationship itself. > (and the probability to loop to itself in the node itself). > > However... it seems (accoring to the relationship api) that i can't > set a property for a link. > Is that right ? > > Any idea about how to store a "relationship property". > Perhaps an index of relationship with "key = relationshipID" and > "value = propability to follow this link" ? > > thank you for your help. > > -- > Laurent "ker2x" Laborde > Sysadmin & DBA at http://www.over-blog.com/ > _______________________________________________ > Neo mailing list > [email protected] > https://lists.neo4j.org/mailman/listinfo/user _______________________________________________ Neo mailing list [email protected] https://lists.neo4j.org/mailman/listinfo/user
