Online Markov Logic Network (MLN) Solver

rockit logo Welcome to the online interface of the Markov logic network solver rockIt. RockIt solves maximum a-posteriori (MAP) queries and marginal queries formulated in Markov logic networks. In rockIt you can define both, deterministic knowledge (hard formulas) and probabilistic knowledge (soft formulas). We refer to for further information.

Run RockIt


There is a documentation including simple installation instructions. RockIt's syntax is almost identical to that of existing Markov logic systems such as Alchemy and Tuffy. The main differences are:

  1. A predicate definition preceded by * is considered to have the closed world assumption; i.e., all ground atoms of this predicate not listed in the evidence are false. (Example: *friends(Human,Human))
  2. In RockIt, we do not need to define query predicates explicitly, because every predicate not preceded by * is a query predicate.
  3. Variable names can also start with a capital letter. Constants must be enclosed with double quotation marks (Example: "Constant").
  4. In RockIt, it is not possible to use implications (=>) and conjunctions (^). The user has to transform her formula to CNF (Conjunctive Normal Form) such that they only contain disjunctions (v).
  5. The syntax of existential formulas differs from the syntax of Tuffy and Alchemy. Please refer to our documentation for details.


You can start building your model from the following example MLNs:

About This Service

Please note that we restricted the time of the ILP solver to 30 Minutes per cutting plane iteration in this online version. After 30 minutes the solver will return the best solution which it currently found. The maximal overall runtime is 3 hours per request. If your model requires more than three hours to be solved, the execution is cancelled.

This service is run on a Dual Core virtual machine with 8 GB ram. Please note that due to resource restrictions we run more than this service on the same machine. Thus, runtimes might vary.

Rest Interface

Programmers can also access our service via rest interface. A description of the required API parameters and a demo client is available here.