Crowdfunding "Sentiment" - Anyone interested? Here is our plan for a feature

Coordinator
Jul 2, 2015 at 5:48 AM
Edited Nov 11, 2015 at 7:14 PM
Hello NodeXL Users!

We are considering a new feature and we would like to ask for your support.

Image

NodeXL has some content analysis features, but many people ask for "sentiment".

Some of us have a very negative sentiment about sentiment, but we do understand the need for a way to measure the orientation of a collection of content.

We have a design for a feature that we think will provide many of the services of a sentiment system with more transparency and flexibility.

We plan to allow NodeXL users to edit the names and contents of three lists of words.

Image

We will provide a default list of American English language "Positive", "Negative" and "Neutral" words. Users can rename and modify these lists.

NodeXL will then count the frequency of the words from each list at the Edge, Vertex, Group, and Network level.

Users could replace the lists of words with similar lists in different languages. Or a completely different set of word groups could be used.

This is a simple way to provide a measure of the orientation or polarity of a message, individual, group, or network.

We imagine this taking several weeks to implement and test.

As we plan our next features we need to assess demand and support for the project.

Is this feature a high priority one for you?

If so, would you donate to support the implementation and public release of the feature?

We'd like to enhance NodeXL in many ways, including time series analysis, diffusion analysis, and better visualization and web interfaces.

Let us know what your priorities are for NodeXL development!

Image

Regards,

Marc
Jul 4, 2015 at 1:35 PM
Hello, Marc: I think a sentiment analysis approach would be helpful. I like the idea of being able to (see and) modify the word lists and where they lie in terms of sentiment. The positive, negative, and neutral are rough categories. Maybe someday there could be more complex analysis, such as "emotion" analysis (as described in the research literature).

The time series analysis, diffusion analysis, and so on, would be cool as well. Nice sense of direction and ambition.
Coordinator
Jul 12, 2015 at 2:35 PM
Thanks for the feedback!

With the support of the community, we plan a series of enhancements designed to summarize and extract insights from "collections of connections" (i.e. "networks").

Regards,

Marc
(For Team NodeXL)