5
$\begingroup$

We have several hundred employees processing word documents, mainly applying some defined rules on how the final document should look like. Some of those rules are: title formating, figure positions, text style among others.

I would like to optimise this process and replace the human by AI. The human will move from processing to quality control (QC) so we will have a much higher quality.

  • My question is: How to approach this problem and what would you recommend as initial start point?

Background: I am new to AI, although I've been following this topic closely for about a year, and want to move forward with implementation. I have experience with VBA, PHP, C, C++, and some other languages.

|improve this question
$\endgroup$
  • $\begingroup$ Welcome to AI! Can I ask, do you mean Microsoft Word documents? (If so, I'd specify. Also, it might be better to use something like "Automated formatting of text documents" for the title. "Treatment" is a little vague, but "Formatting" is a good umbrella descriptor for what you want to do.) $\endgroup$ – DukeZhou♦ Sep 11 '17 at 16:19
  • 1
    $\begingroup$ Hi Duke, you're right, it is about MS Word. I edited the title as suggested, thank you. $\endgroup$ – Miloš Aug 23 '18 at 7:41
  • $\begingroup$ What have you done so far? Even though you might sound as a new contributor,but as the rules flow expert answers to question problems,keeping in mind,that being newbie doesn't save your project question from failing. So what have you done so far? So as we can start from $\endgroup$ – quintumnia Aug 23 '18 at 18:58
  • $\begingroup$ @quintumnia, thanks for the comment. So far I was learning about AI and Machine learning, theoretical part only. As this is for now a side project, I do not spend a lot of time on it. The starting point as described in the question is: "How to now link Machine Learning and text processing". Which approach can be used in order to be able to identify, parse and apply some text formatting following specific rules and in the meantime apply machine learning. My apologies if the question is confusing as I'm far from being expert on this topic. $\endgroup$ – Miloš Aug 24 '18 at 10:32
1
$\begingroup$

Word documents, processed by employees in a company are discussed under the topic “Workflow management system”. That means the process is fixed and the question is, which person is doing what. Optimizing the process is easy, it is called outsourcing which means to transfer the whole business process to the Philippines because the average income there is lower and the local population speaks English very well.

A second part of your question is about using AI to improve the workflow. That is not possible with today's technology. A “Workflow management system” is the opposite of automation, it means that human workers are doing the job, and that human workers are communicating over the result of their work. What you can do is to automate the communication process in the group. The standard tool right now is e-mail which is send over the intranet. Automating the communication would need chatbots on human level which are not available right now, that means the gap to human knowledge is too big.

From an abstract point of view, the business process in a company until a word document is finished including the layout and placing of figures is the result of a workflow. All workflows are driven by humans. Improving the quality of existing workflows can done with:

  • better management
  • outsourcing
  • faster communication

But the workflow remains the same. That means it is not automation it is group working. What i want to express is that modern programming languages like VBA, C++ and PHP are not helpful.

|improve this answer
$\endgroup$
  • $\begingroup$ Hi Manuel, sorry to say this but I really think your answer is not constructive at all. Humans are going to moon since long time now, many advanced things are possible. I have discussed with some researchers working on NLP (Natural Language Processing) and doing some really great staff. I'm sure something can be done on the formatting and layout part of a word file. Note that a word file can be converted to html, which makes it's formatting very easy. $\endgroup$ – Miloš Aug 27 '18 at 7:08
  • $\begingroup$ @Milos Since the 1960s, many AI related projects were started with the aim that computers can create and modify documents for themself. The CALO project (a cognitive assistant) is known for his capabilities in speech recognition (iphone SIRI is the commercial version). In the literature the term “automatic document formatting” is used for describing systems which can generate Word documents autonomously. $\endgroup$ – Manuel Rodriguez Aug 27 '18 at 8:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.