For many languages there are no large, general-language corpora available. Until the web, all but the richest institutions could do little but shake their heads in dismay as corpus-building was long, slow and expensive. But with the advent of the Web it can be highly automated and thereby fast and inexpensive. We have developed a `corpus factory' where we build large corpora. In this paper we d...

Word sketches are one-page automatic, corpus-based summaries of a word’s grammatical and collocational behaviour. These are widely used for studying a language and in lexicography. Sketch Engine is a leading corpus tool which takes as input a corpus and generates word sketches for the words of that language. It also generates a thesaurus and ‘sketch differences’, which specify similarities and ...

Editing a sketch should be one of the essential features provided by sketch recognition systems to allow people to modify what they have drawn, without having to delete and redraw shapes. This paper introduces a control point based editing approach we call RingEdit. RingEdit differs from other sketch editors in that the user actually draws their own control points on the sketch, rather than rel...

Handling an Update: When an update (it, ct) arrives, then ct is added to one entry in each row of the array count. Specifically, ∀1 ≤ j ≤ d, count [j, hj(it)]← count [j, hj(it)] + ct. Lemma 1. The space used by Count-Min Sketch is O(wd) ≡ O( ln 1 δ ) words. Specifically, it uses an array which takes wd words and d hash functions, each of which can be stored using 2 words. An update can be handl...

We study a natural generalization of the heavy hitters problem in the streaming context. We term this generalization block heavy hitters and define it as follows. We are to stream over a matrix A, and report all rows that are heavy, where a row is heavy if its `1-norm is at least φ fraction of the `1 norm of the entire matrix A. In comparison, in the standard heavy hitters problem, we are requi...

A part of the case for corpus building is always that the corpus will have many users and uses. For that, it must be easy to use. A tool and web service that makes it easy is the Sketch Engine. It is commercial, but this can be advantageous: it means that the costs and maintenance of the service are taken care of. All parties stand to gain: the resource developers both have their resource showc...

We use Abstract 3D models for sketch recognition. The models should be abstract because sketches do not contain texture information which is present in the usual 3D models. Zernike moment descriptors are used for recognition. These descriptors do rotation invariant recognition and also provide the angle of rotation. Also the same concept can be extended to 3D viewpoint detection of the sketch.

The creation of 3D computer models is an essential task for many applications in science, engineering and arts and is frequently performed by untrained users. In many cases speed and simplicity of the modeling process is more important than matching the geometry of the modeled object exactly. Sketch-based modeling has been suggested as an important tool for such applications. In this paper we e...

Two tasks are prevalent in most sketch recognition systems grouping and recognition. We describe a novel grouping algorithm which clusters pen strokes into symbols. This algorithm is unique because it allows a feedback loop which can improve the performance of other phases in recognition. In our experiments, we improve the percentage of perfectly grouped symbols from 76.7% to 94.2%. Additionall...