Posted by Timne on July 9, 2009, at 11:56:46
This is not available yet on source forge and might not ever be.
In reading posts here, I notice that many people get engaged in conflicts - including the forum provider -- when they don't agree on proper language for conveying messages here. I suspect many mental health professionals will agree that often the problem is not one of the posters' intent, but rather one of habit, and the lack of knowledge that would make other habits more accessible. We can make a machine for that.
With the ability of JavaScripts to make asynchronous server calls, such as the behind the scenes scripting that offers proposed search terms in the Google search box, technology could assist posters to comply with the host's conduct demands.
In concept, the algorithm seems simple. It could start with two manually compiled lists -- flagged word combinations and proposed alternates. As the lists accumulate actual suspect combinations, the combos can by classified by type (e.g. 2ndPersonNoun + ToBeVerb). Integration of the lists in a dynamic spell-check or grammar check format could follow.
The statement checker would first scan for word combinations involving second or third person nouns. It might search by type of combination, as in the above example and for actual combinations ("You are..."). "To be" verbs would top the list of suspect pairs. For example, "You are..." would be a definite flag. "She is..." would be a flag.
On the other list, we have replacement word combinations. "One way to look at it is that a person is..." would be offered as a replacement for "She is..." I would call that a high-obfuscation phrase, which from my experience is the main substance of the "I statement..." so preferred by some clinicians. But if they say it works, we can make a machine to show generally how it's done.
Further down the list, beneath the high-obfuscation phrase, we could find somewhat less obfuscating phrases such as "A person is..." or "One is..." The machine would never know the right answer, but it could offer alternatives one could use in one's own analysis.
I don't suggest this would be an effective means of enforcing the site host's preference, and I believe that to leverage it as an enforcement tool could contribute to more strict and conflict-generating enforcement of concepts that don't seem well understood here already. It's more like a training aid, and an automated helper for writing for a distinct venue. But, if the P-Doc is convinced his rules here have therapuetic value, a machine that helps people appreciate the substance of his expectations would seem an effective technology to deliver that therapeutic value.
Am I daydreaming what might be possible, or am I qualified to produce this?
var Posted = form.post;
var badStringList = httpRequest(AJAX call to list of suspect word combinations);
var re = new RegExp(document.badStringList.regex.value);
if (document.Post.subject.value.match(re)) {
return badstringlist;
} else {
alert("Your post contains no suspect word combinations currently on the suspect list");
}
}Apparently not, or at least I'm not demonstrating working code here. But neither am I suggesting that the P-Doc here has the programming skills, time or where-with-all to implement this or that he should. I'm saying, with the full authority of an anonymous person on the Internet, that it could be done.
You -- nay -- (One) probably get(s) the idea, if you(one) know(s) a bit of JavaScript, that my hasty requirements list is somewhat authoritative. A regular expression relies on an AJAX call to a server-side script to get a list of word pairs. That server script relies on both compiled lists and dictionary lists to identify second- and third-person nouns paired with "to-be" verbs or other suspect verbs, maybe commonly used strong adjectives, and searches the post for the presence of those combinations.
If matches are found, the browser calls the server again to construct a list of proposed replacement terms, which it returns to the browser that in turn offers those words in a list much the way Mozilla offers alternate spellings if one right-clicks on a term that is highlighted as a spelling error in a text field.
So, yeh, maybe I can do it. And it makes sense to me that if we are invited to extend our rhetorical involvement into new and sometimes troubling machine-facilitated dialogue, that we also might rely on machines to help us construct dialogue to fit particular contextual expectations.
poster:Timne
thread:905793
URL: http://www.dr-bob.org/babble/admin/20090707/msgs/905793.html