is interpreting information possible in Access 2007?

[s]

I have a table which is populated with fields as

Year | Department number | Department Contact Person | Issues Faced |
Steps taken to resolve them | Satisfaction of workers in department

The values of the records are

2005 | 1000 | John | Lack of motivation among staff | Conducted weekly
meetings with staff and so on | It has increased to 5% from last year.
There is less turn over and so on.

2006 | 1000 | John | Lack of motivation among staff | Encouraged group
projects with staff | Satisfaction amongst workers has increased from
5% from last year to 9%.

The ' | ' is used to differentiate between fields while posting

Question is can Access interpret that for me if I want to know what
was the improvement in "Satisfaction of workers in department" for
department number 1000 in the year 2006 from 2005. With my limited
knowledge of Access, I don't think so, but can someone please clarify?

Is there a way, I can have Access do that using some VBA code or will
I have to delve into natural language processing a lot for this(like
http://sites.google.com/site/yaoziyuan/ideas) which cannot be done in
Access? If the human language description can be converted into
something which can be manipulated by Access for analysis, that would
save me a lot of work of manually interpreting it.

I would appreciate any suggestions and advice. I wonder how others
solve this. Do any other RDBMS packages offer the advantages of the
interpretation I need?

Thanks

 

[The Frog]

Hi there,

The only way that I could think of that you could do something like
that would be to doing some string processing on the 'Issue Faced'
field, looking for key words, and generating a positive or negative
value for the 'key' words that are in the field. The 'worse' or
'better' the word the greater the value given to it in either a
positive or negative value - then sum them up at the end -> gives you
an overall rating for the field.

Its not perfect by any means, but it might allow you to get close to
what you want.

You would need to build a 'dictionary' table with ratings assigned to
each word, say for example from -10 to 0 (neutral) to +10 - giving you
21 possible ratings for each word. You would then run a process to
collect each individual word from each records 'Issue Faced' field,
check them against known words in the dictionary, and if they are
unknown present them in a form or something similar for you to rate
them. Once all the words are known and rated (which could take a while
but probably only really needs one major hurdle at the start and
little ones in the future), you could run another process to calculate
the 'rating' for each record, store it somewhere (another field
maybe).

Now you have the information you want, so its up to you to decide how
you want to display it, analyse it, etc....

Another way would be to simply have a set of fields to capture the
required rating information as people enter the data. But that wont
help you for the historical data.

Food for thought

The Frog

 

[s]

Thanks to both posters for their advice and time.

However, is there a way natural language sentences like "There is less
turn over and so on."
could be interpreted by Access 2007 in some way? Some of the records
have lot of
sentences like this. I realize it is part of natural language
processing
and don't know how many RDBMS packages provide that so would
appreciate any advice on how to interpret such
statements.

Thanks again to both posters
for their aid and time.

 

[Banana]

However, is there a way natural language sentences like "There is less
turn over and so on."
could be interpreted by Access 2007 in some way?

Parsing natural language is going to be a major undertaking.

If it were up to me, I'd just look somewhere for a library thta provides
this functionality. If it's free, happy days. If it charges money, well,
bill it to the client and see if they really want that functionality.

It's also possible that FULLTEXT search may be good enough, in which
case you would want to use another RDBMS as a backend, such as SQL
Server, MySQL, Oracle, or whatever that has the FULLTEXT functionality.

 

[s]

Parsing natural language is going to be a major undertaking.

If it were up to me, I'd just look somewhere for a library thta provides
this functionality. If it's free, happy days. If it charges money, well,
bill it to the client and see if they really want that functionality.

A quick search does not reveal any library providing such
functionality

It's also possible that FULLTEXT search may be good enough, in which
case you would want to use another RDBMS as a backend, such as SQL
Server, MySQL, Oracle, or whatever that has the FULLTEXT functionality.

Yes, I am thinking of using MySQL for that part. Full text may help
us
to some extent. I hope some such library exists for those RDBMS
products.

I appreciate your and the other person's advice and time.

Thanks
Thanks