This morning’s lecture gave a general overview of statistics and their role in analysing quantities of data. Most of the technical constructions — mean, median, mode, standard deviation — are probably familiar to many, but the setting for their application and the computational context may not be.
Continue reading Lecture 17: Summary Statistics
Today’s lecture presented various techniques to support effective information retrieval: the big bag of words model; term-frequency inverse document frequency (tf-idf); the vector space model; and cosine similarity for document ranking.
Continue reading Lecture 16: Vector Spaces for Information Retrieval
In the most recent tutorial exercises you used the
cqp tool to search a 3-gigaword Dickens corpus. We also have the 96-gigaword British National Corpus installed under
cqp which you can explore by selecting
BNC at the commmand line.
$ cqp -e [no corpus]> BNC BNC> AllWords = [word="[a-zA-Z].*"] BNC> size AllWords 96063265 BNC>
This has part-of-speech and lemma information like the Dickens corpus, using the Claws 5 POS tag set. As this corpus is much larger you will find queries take noticeably longer to execute.
I also recommend reading the following article on the design and creation of the BNC.
- Gavin Burnage and Glynis Baguley. The British National Corpus. Library and Information Briefings 65, February 1996.
This includes information about text corpora in general, as well as specific details about how the BNC came about.
The tutorials web page now has the latest set of tutorial exercises. The coursework assignment has also been running for a week now, and is due in on Thursday next week. The Information Retrieval tutorial work is fairly brief, which will allow you to spend time in the tutorial discussing any questions you have about the assignment. Please do take advantage of this: attempt every question in the assignment before your tutorial, and note down any concerns you have.
Following the rectangular tables of relational databases and the triangular trees of semistructured data, the remaining Inf1-DA lectures will address the representation and analysis of more unstructured data. Today’s lecture provided a brief introduction to the classic information retrieval task of searching a large collection of documents to find those that match a simple query.
Continue reading Lecture 15: Information Retrieval
Corpora are widely used for computational research into language, and for engineering natural-language computer systems. In linguistics, they make it possible to do real experimental science: to formulate hypotheses about the structure of languages, or changes in language between different places, times or people; and then test these on data. In building applications that handle text or speech, corpora provide the mass quantities of raw material used for machine learning and other algorithms.
Continue reading Lecture 14: Example Corpora Applications
The tutorials web page now has notes and solutions for Tutorial 5 as well as exercises for Tutorial 6
Continue reading Tutorials: Notes and Solutions, New Exercises