We are happy to announce the first release of TopicExplorer 1.0, Japanese Blog-Analysis-Distribution that includes general features for browsing topics and related documents, temporal analysis and topic frames.
General Features
- Browse topics in serialized order with respect to topic word similarity
- Represent topics by relevant words
- Rank documents with respect to relevance to a given topic
- Show topic assignments of tokens within a document
- Topics are always referenced by color and tool tips
Time
- Show number of tokens assigned to a given topic over time
- Allow comparisons to token time series of other topics and to the average
- Show relevant words of topics for particular weeks
- Rank documents of a particular week with respect to relevance to a given topic
- Toggle rankings between relevance and temporal ordering
- Download time series data as CSV-Export (Format of CSV still needs to be specified)
- Show always date and time of documents
Co-locations (Frames)
- Define a topic specific co-location (frame) as combination of verb and noun tokens both assigned to the particular topic and occurring within 150 tokens in a document and are not interrupted by dots and other syntactic characters. Current list of frame interrupting characters: 。, ？, ?, ！, !, ．．, .., ・・, ．．．, …, …, ・・・
- Represent topics by frequent co-locations of verb-noon combinations (frames) and show number of documents contains a particular frame
- Show total number of frames and unique frames of a particular topic
- Rank documents containing a certain co-location with respect to relevance to a given topic
- Show frames within a document
