Bonjour,

Programmeur il y a longtemps, il y a bien des choses que je ne sais plus faire ... surtout quand c'est urgent !

Le problème :
Je traite une grande quantité de données textuelles (données issues d'un forum, environ 2900 post, en anglais) pour faire une analyse qualitative.
Les outils d'analyse qualitative que j'utilise ne gèrent pas les stopwords (les mots vides , comme dans ce texte les je les ne j' pas ...)

J'ai recherché différents programme ou scripts permettant de faire le travail ... je n'ai rien trouvé sauf ici : http://maya.cs.depaul.edu/~classes/c...ocess/stopper/

Le programme fonctionne correctement (il y a une petite erreur dans strlist.h) mais il supprime tous les sauts de ligne, ponctuation, etc ce qui fait que je me retrouve avec une liste de mots. Je n'ai plus mes paragraphes.

un exemple en entrée :
Code : Sélectionner tout - Visualiser dans une fenêtre à part
1
2
3
4
5
6
7
8
9
10
11
12
Project Management Implementation is a full time job and needs to be handled as such.
Based on recent Harvard reports and comments thereon, seems like everything is wrong with the EMR implementations - be it people, processes and technology.
I have also read a few reports that state failures of EMR/EHR implementations, but they lack specifics as to what caused these failures. Are the requirements not explicit or understood correctly? Are the business processes mapped correctly? Are there functional gaps? Is it an issue of inadequate training or too complex a software?
"No 1. - Lack of Governance
PS. I do not see that legislation per se is detrimental to adoption of EMR/EHR, and as a matter of fact see it as being crucial as part of the overall EMR/EHR bigger picture."
Each Provider/Speciality has unique work flow. Physicians tend to adopt different techniques in gathering symptoms and approaches to diagnosis all seems to achieve the objective. Thus customization becomes too complex.
"A couple of thoughts, in no particular order:
- Healthcare is often reactive, rather than proactive, when it comes to opportunities for improvement. It is so often a race to fix what went wrong, not institute infrastructure to prevent it. Have to run to find the code cart; have to get a patch to fix that problem with accessing the data warehouse. Not planning the optimal place for the code cart; not planning the best long term strategy for IT.
It is an issue of trust, and the way that healthcare seems most frequently to grant trust is by seniority rather than by relevant experience and aptitude. I really wish I knew how to change this, as it is really problematic. Too many smart, interested people who see the potential of software technology get turned away. Unfortunately, I am going to have to agree with you, John, that it probably will take some (many) pains of failure before advances can be achieved. As well, it is a matter of how long the country will tolerate this, and when the right incentives are put into play to make administrators see that leadership change in this area is much needed."
"Another possibility is the dearth of skilled managers in healthcare.
All of us are, of course, excepted from this description, but all of my classmates wanted to go to Goldman Sachs. If they had an interest in healthcare, they went to Genentech. Having the fiefdom oriented departmental management, precluding the ability to communicate across these artificial boundaries, and just having excess myopia throughout these organizations means that there is little understanding of the business within many healthcare providers, Successful implementations usually have some aspects of Business Process Management, a clear understanding of care provision in a technology enabled environment, and the ability to accurately measure and achieve the improvements in captured contribution margin before, during, and after implementation. It certainly does not help that most of the companies that sell HIT solutions are designed and delivered in the antique version, some 30 - 40 year old designs with some front end adjustments. They get away with it simply because much of their market does not know the difference."
"Doctor and nurses are tired folks... Asking them to enter data while seeing patients is little too much.
me donne cela
Code : Sélectionner tout - Visualiser dans une fenêtre à part
project management implementation time job handled based recent harvard reports comments thereon wrong emr implementations people processes technology read reports failures emr ehr implementations lack specifics caused failures requirements explicit understood correctly business processes mapped correctly functional gaps issue inadequate training complex software lack governance ps legislation se detrimental adoption emr ehr matter crucial overall emr ehr bigger picture provider speciality unique flow physicians tend adopt techniques gathering symptoms approaches diagnosis achieve objective customization complex couple particular healthcare reactive proactive comes opportunities improvement race fix wrong institute infrastructure prevent run code cart patch fix accessing data warehouse planning optimal code cart planning term strategy issue trust healthcare frequently grant trust seniority relevant experience aptitude wish change problematic smart people potential software technology unfortunately am agree john probably pains failure advances achieved matter country tolerate incentives play administrators leadership change possibility dearth skilled managers healthcare course excepted description classmates goldman sachs healthcare genentech fiefdom oriented departmental management precluding ability communicate artificial boundaries excess myopia throughout organizations means little understanding business healthcare providers successful implementations usually aspects business process management understanding care provision technology enabled environment ability accurately measure achieve improvements captured contribution margin implementation help companies sell hit solutions designed delivered antique version designs front adjustments simply market difference doctor nurses tired folks enter data seeing patients little
Au lieu de quelquechose qui pourrait ressembler à cela
Code : Sélectionner tout - Visualiser dans une fenêtre à part
1
2
3
4
5
6
7
8
9
10
11
12
project management implementation time job handled.
Based recent harvard reports comments thereon wrong emr implementations people processes technology
read reports failures emr ehr implementations lack specifics caused failures requirements explicit understood correctly business processes mapped correctly functional gaps issue inadequate training complex software
lack governance
ps legislation se detrimental adoption emr ehr matter crucial overall emr ehr bigger picture
provider speciality unique flow physicians tend adopt techniques gathering symptoms approaches diagnosis achieve objective customization complex 
couple particular
healthcare reactive proactive comes opportunities improvement race fix wrong institute infrastructure prevent run code cart patch fix accessing data warehouse planning optimal code cart planning term strategy
issue trust healthcare frequently grant trust seniority relevant experience aptitude wish change problematic smart people potential software technology unfortunately am agree john probably pains failure advances achieved matter country tolerate incentives play administrators leadership change
possibility dearth skilled managers healthcare
course excepted description classmates goldman sachs healthcare genentech fiefdom oriented departmental management precluding ability communicate artificial boundaries excess myopia throughout organizations means little understanding business healthcare providers successful implementations usually aspects business process management understanding care provision technology enabled environment ability accurately measure achieve improvements captured contribution margin implementation help companies sell hit solutions designed delivered antique version designs front adjustments simply market difference
doctor nurses tired folks enter data seeing patients little
Quelqu'un pourrait-il regarder le code de façon à conserver, une fois modifié, la structure des paragraphes, stopwords en moins

je vous en suis très reconnaissant

Bien à vous

TGM