I'm a BA in Math/CS with ~20 years' experience in Unix system management and development. I'm now back at school to get higher degree in History.
I'm looking for ideas for a thesis that will combine History and Computer Science (e.g., create a computerized application that will help historians, and is something new).
Anyone has any ideas?
How about a visualization that compares on a timeline research papers versus patent applications? That would be a good start.
if you want to help the historians with a new application may be a good start would be to ask them about what would improve their work and what makes it difficult. Probably there is an history feed on RG
Mobile app which gives fast result on current locations History (based on GPS) + info about ppl who made history in those locations..
if u r devolepin an app let me.. I have lot of time to kill now :)
can you explain which history you actullay focusing on?
Once type of history is like news events , political history.. ? or you talking about how people use digital documents to complete their assignments which include simple activity like booking for a hotel to writing a thesis on a topic?
1.The google release of n-grams http://ngrams.googlelabs.com/ is a simple but powerful tool for examining the frequency of sets of words. I am sure it could enhance much historical study.
2.I am interested in the NLTK for python which is a powerful library for anything to do with Natural Language Processing, recommended especially if you use python.
3.I also have a little dream of making "social simulations" using ideas like Game Theory. The simplest sim I know is wa-tor ( http://www.leinweb.com/snackbar/wator/ ) it's a stab in the dark but I wonder if the use of resources in a human society could be modelled simply. Maybe see how oil use varies in a sim or something.
4.@Kees I like the idea of attractors that govern human social processes. I wondered about the attractor that governs the interplay of all neurotransmitters in the brain, it would be a complex one and hard to visualise.
@Ron thanks for the challenge
Thanks Luke Dunn for your excellent pointers.
I will educate myself on Google's n-grams (http://ngrams.googlelabs.com/) and NLTK for python (yes, I do script in Python).
I once had the notion of creating a logical (as opposed to statistical like Google translate) parser for Latin, as it's more structured than other languages. After examining the obstacles
of parsing idioms and having to analyse whole paragraphs to get the correct meaning of a certain phrase, I determined that the task is too dificult (ignoring something like IBM's Watson (www.ibm.com/innovation/us/watson/index.html)).
However, I'll look at the project you mentioned and see if it may be of help in building such a parser.
According to me, i am search about an idea for my master thesis in infomatics, any one can help me ,please.
Make a program which analyse past events and makes inferences whether humanity learns from it's mistakes and which mistakes are repeated and etc. In conclution - Make a humanity stupidity analyser.
@Ron I was an unquestioning recipient of the (last days of the?) obligatory classical education when at school here in the UK. I still wonder at Latin's terseness and elegance. But yes, pragmatics is to me still the biggest problem with NLU, the human mind seems to latch onto a fine web of many contextual clues when it manages to disambiguate between different meanings for something like "Time flies like an arrow"... my favourite interpretation of this is "Oh how time speedeth by, d'ya fancy an arrow ? (there's a darts board over there by the bar)" followed by "(the unusual african insect) the 'time fly' is universally found to form a fondness for arrows."
Hi Artificial Neural Networks and Statistics applications can help to your idea, Reason : with these applications you can estimate the behavior unknow system with existing one.
Here is a nice example, I believe that Diana had something very similar in mind: https://www.recordedfuture.com/index.html
Below is a preliminary description of the tool that will be the main part of the thesis I envision.
o From your areas of expertise - could you suggest which research areas I'd need to educate myself on - to be able to create this tool?
o Which tools do you know which may used as "building blocks" for parts of the tool?
--------------------------------------------------------------
General:
The thesis will include the creation of (as automated as possible) tool, that will get as input a textual historical source (or sources) describing an event or phenomenon. The tool will break the information contained in the sources into its fundamental data units (memes), and construct a data structure from them. It will then apply heuristics on the information to identify patterns that may aid in deriving conclusions about the event or phenomenon.
Details of the tool:
o The tool will be constructed of interacting modules. That way, some of the modules could be implemented as manual (non-computerised) in case the state-of-the-art isn't up to par to automate them.
For example, candidates for non-computerised modules are the breaking down of the sources into memes, and populating the data structure.
o Data entering - there may be a need for an intermediate step of natural language translation into formal language, when transposing the source (that is in a natural language like Latin, Greek, or English) into the data structure.
o The superstructure of the tool will be in a high-level scripting language (e.g. Python) for convenience and rapid development.
If necessary - optimization of bottlenecks discovered by profiling, will be done in lower level languages (like C).
Tool's requirements:
o The tool should be able to analyse the data structure to find statistically significant patterns.
o The tool should be able to describe to the user how it reached its conclusions (i.e. the tool will not be a "black box").
o The tool should detect that two memes are identical (or at least similar. I need to define when two memes are similar enough to be considered identical).
Questions to ask myself:
What criteria determine that two descriptions (e.g., from two historical sources) refer to the same meme?
For example, in case of a certain war – the following data should be compared: who were the fighting forces, who were the armies' commanders, dates, geographical regions, the season of year, the duration of fighting...
Have to think about:
o Which parts of the tool already exist (so no wheels are re-invented)?
o How to give sufficient flexibility to data records - so more links and fields could later be added?
o What should the data structure be?
o Which links are needed among records and fields?
o selection of a knowledge representation formalism
o selection of a mining algorithm
o selection of an analysis algorithm that would build this representation
Case Study:
The test case for the tool will be the Military Tribune with Consular Authority magistracy, that existed between 444 and 367 BC, in the Roman Republic (Consular Tribunes).
(Historical background: In most years of the Roman Republic, the highest magistrates were the two Consuls. However, in the above period of around 70 years, the Roman popular assemblies would sometime forgo Consuls and elect Consular Tribunes. Unlike the Consuls, they could be plebs, not patricians and their number was more than two).
Other possible test cases:
o legend of Hercules at Rome
o Abraham and Heracles
Questions I would want to answer (or at least offer hypnoses) regarding Consular Tribunes:
o Why were Consuls elected on a particular year and on other years Consular Tribunes?
o Why was a particular number of Consular Tribunes elected on a particular year?
o Why were plebes/patricians elected on a certain year?
Already existing building blocks:
o TBD
Possible applicable research domains:
o Information Extraction (creating metadata)
o Finite Automata
FASTUS (MUC-5)
o Context-free Parsers
o Named entity extraction
o Topic detection
o MUC - Message Understanding Conferences (DARPA)
MUC-6
o Tools:
GATE (http://gate.ac.uk/)
o Question Answering
o Chatbots (Eliza)
o Textual Entailment
o Text Mining
o Text Analysis:
o "NLP toolkit"
o Stanford NLP tools (http://nlp.stanford.edu/software/).
o Classification:
o WEKA.
o Parsers and Grammars:
o Penn Treebank Project
o Context-free Dammara:
History Assistant
CYK algorithm
o No category yet:
o Thesauri: WordNet
o Conceptual indexing
o SIGIR
Examples of historical (OO) classes:
o Nation
o Tribe
o Family
o Individual
o Event
o Alliance
o Historical Source
Examples of fields in historical (OO) classes:
o Description of event
o Date
o Protagonists
o (Power) interaction
o Belonging to which nation/party/social class?
o Motivation
o Start (initial) conditions
o End (final) condition
o Duration (of life, battle, war, existence)
o Which were the source
o How reliable (biased) is the source
Hi Ron, I wish you success in your project.
I am a little confused by your iintention and need a clarification. Are you seeking a degree in a History department or a computer science department ? I am asking this question because on the one hand you said " a higher degree in History". On the other hand, all the conversation is orbiting around the creation of a tool that a student in a computer science department would design.
You can use data mining to search for hidden connection between different history events that are not relative apparently. I hope it is helpful to you.
Answer to Mamadou Kone
This Thesis is rooted, as you noted, in both History and CS. My intention is to first create the system (the CS part of the thesis) and then use this system to try and find patterns in historical issues.
In the end, I think the degree would be in History, as I'm trying to use as many existing "building blocks" as possible, and assemble them (while writing the needed "glue" parts).
Hello why don't you use data warehousing to retrieve the old history data and analysis for new trend or new generation people. If you interest in market so many open source software's are available. I hope you can concentrate on this data warehousing also.
Answer to Bhaskar B
The main problem, as I see it, in the data acquiring - is to actually get he data off the historical sources. The storing of data later will be in the format most convenient to apply heuristics to it and discover patters.
Are you sure data warehousing will help in extracting the facts from the texts?
Can you suggest any data warehousing products that know how to extract data from literary texts?
Hi Ron Barak,
We can do the data acquiring, data cleansing, data mining, data scrubbing and etc in the part of the data warehousing. In data warehousing we have ETL software’s. Through ETL software we will take different type as a source data (like flat files, text files, CSV files, relational and dimensional data sources) and extract , transform and loading into data warehousing. .
Have you looked at Google's Ngram viewer - it's very cool and might give you some ideas - http://ngrams.googlelabs.com/
Have you looked at Chronozoom - it may spur some ideas. Very Cool, but i am a bit biased :)
http://eps.berkeley.edu/~saekow/chronozoom/
"ChronoZoom is a step toward enabling users to browse knowledge, rather than searching, by linking relevant online resources to fixed time scales in an organized and logical manner."
This may not be exactly what you're looking for, but if you want some general inspiration about the overlap between computer science and history, try reading Neal Stephenson's Cryptonomicon / System of the World books.
A search engine that filters out non-historical sources, shopping sites or forums for example, leaving only original sources or enthusiast sites.
In answer to Bhaskar B:
Can the ETL software deal with historical texts, automatically?
Wouldn't the Extract and Transform need to be done manually, by a human.
For instance, if the text is http://www.perseus.tufts.edu/hopper/text?doc=Perseus:abo:phi,0914,00123:35 , is there an ETL software currently capable to extract all the facts from that text? For instance, the text mentions the "battle of Cannae", will the ETS software know what a battle is, could it deduce from the text (and other related texts) who the participants in the battle were, what were the preconditions to the battle, what were the battle results (e.g., who won), the number of casualties, the length of fighting, etc?
Hi Ron,
You said in your reply to my inquiry : "In the end, I think the degree would be in History". I guess you mean that it is the history department (or the like) that is going to grant you a degree. This looks very much like an integrated degree. Besides, I am curious to know how much the folks in the history department are going to understand the core of your system. This is an original work in an unusal context.
I wish you all the best.
Hi, in Poland we have PTI - Polish Informatics Association which arrange dynamic activity in IT history - especially in Poland, but not only. If you want to contact with this people I give you two addresses:
'Jurek Nowak' [email protected] , Piotr W. Fuglewicz [email protected].
Great part of Polish informatics interested in IT history live out of Poland – in USA, New Zeland, Western Europe and so on. I hope that you will find their support.
Rds
Have you explore the cultural computing. This is a combination of history and computer science and latest unexplored area in research.
yeah, why don't you build a simulated ww2 Germany navy 'shark enigma machine' application, get the graphics right and it would be really cool. people could send and receive U_boat instructions. just like the real thing, maybe even with a 'colossus' British decryption element as well .
history and computer science right there.
I can see...you has to make some more of research about the units of measurements applied of created for social/history sciences, if you have the units, you can decide what project...if the units of measurement doesn´t exitst, you can create it...and how muy effort you want to dedicate to this project? is only for graduate and go to seek a job? pleasure? want to publish? what is your purpose...?
When Ron Barak first posed this question, there were a number of interesting suggestions on combining graduate studies of history and IT. As a follow-up, I'm wondering if there is any update on whether a topic has been identified.
Hi Ron,
Text analytics software's(Tools) currently capable to extract all the facts from that text.
Text analysis software enables users to determine the frequency with unstructured data.
as my suggest we can use text analytics software's(Tools) for intigrating history data to our softwares.then user will get benfit from these history data.
and also this will more use full to society and new generation.
Thanks.
- Badges
- Science topic
- Similar topics
- Computer Science