Review of dissertation proposal
Proposal title:
PhD candidate:
Building the ultimate math search engine
Martin Líška
Proposal content:
The subject of dissertation proposal is the design and development of retrieval system for
mathematic expressions (math information retrieval - MIR). Unlike traditional text retrieval
tasks, the MIR is much more difficult, as the search problem is not just syntactical matching
of terms but more a semantic reasoning task. The author surveys the current state of the art in
MIR, summarizes his previous results and proposes the goals for his PhD thesis. The goals
include semantic basd search (e.g., simplification of expressions, applying algebraic rules),
subformula unification (wildcards for subexpressions), query expansion and combinations of
various approaches.
The proposal is well structured and written in very good English, in fact, it could be used as a
concise survey on MIR. I strongly appreciate the topic, as MIR is still underestimated branch
of information retrieval because of the limited (but still large) number of users (scientists and
professionals applying math). From my personal professional experience, the search engine
for mathematics is highly useful tool especially for interdisciplinary research, where a
researcher needs to model certain problem but is unable to prove/verify the mathematics
he/she developed. The MIR could help identify documents (the math included, respectively)
from different areas of research using similar math; possibly including the stuff missing in the
user's research.
The goals presented in the proposal are very ambitious, although risky. Especially the
semantic search and subformula unification virtually make the search process extremely
complex. On the other hand, as the candidate begun his research already as a bachelor student
ans so is skilled in MIR tasks, the assumptions on finishing the goals are quite realistic in two
years as outlined.
The publication record of the candidate is sufficient, while I expect other will follow
before the thesis submission. I recommend to submit papers to the top IR conferences listed in
the proposal and also in a respected journal.
Questions:
What kind of semantic search do you plan to include in order to keep the search process
efficient?
Conclusion:
In summary, the PhD candidate proved he is able to submit an excellent PhD thesis in a
foreseeable time (2 years). The thesis proposal gives clearly defined goals and presents
original results published in conference proceedings. The candidate's research record is
sufficient, the following years give enough time to disseminate the results prior to the actual
PhD thesis submission. I definitely recommend the proposal to be accepted as dissertation
proposal, as well as the rigorous thesis.
Review:
Masarykova univerzita
In Prague, 12th of March, 20ßakulta informatiky
Datum:
C.jTE.č.:
Počst listů dokumentu:
Počat příloh a listú/sv.:
Počat a druh nejíst pffloh:
J.
J.